EP1253496B1 - Dispositif de pliage comprenant un système de contrôle servant à générer et à exécuter un plan de pliage de tôles métalliques - Google Patents

Dispositif de pliage comprenant un système de contrôle servant à générer et à exécuter un plan de pliage de tôles métalliques Download PDF

Info

Publication number
EP1253496B1
EP1253496B1 EP02002809A EP02002809A EP1253496B1 EP 1253496 B1 EP1253496 B1 EP 1253496B1 EP 02002809 A EP02002809 A EP 02002809A EP 02002809 A EP02002809 A EP 02002809A EP 1253496 B1 EP1253496 B1 EP 1253496B1
Authority
EP
European Patent Office
Prior art keywords
bend
workpiece
bends
sequence
bending apparatus
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
EP02002809A
Other languages
German (de)
English (en)
Other versions
EP1253496A1 (fr
Inventor
David Alan Bourne
Duane Thomas Williams
Kyoung Hung Kim
Sivaraj Sivarama Krishnan
Kensuke U.S. Amada Ltd. Hazama
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Amada Co Ltd
US Amada Ltd
Original Assignee
Amada Co Ltd
US Amada Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Amada Co Ltd, US Amada Ltd filed Critical Amada Co Ltd
Priority to EP06009075A priority Critical patent/EP1684140B1/fr
Priority to EP06002177A priority patent/EP1657607B1/fr
Priority to EP06009079A priority patent/EP1681607B1/fr
Priority claimed from EP95936762A external-priority patent/EP0744046B1/fr
Publication of EP1253496A1 publication Critical patent/EP1253496A1/fr
Application granted granted Critical
Publication of EP1253496B1 publication Critical patent/EP1253496B1/fr
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
    • B21D—WORKING OR PROCESSING OF SHEET METAL OR METAL TUBES, RODS OR PROFILES WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
    • B21D37/00—Tools as parts of machines covered by this subclass
    • B21D37/14—Particular arrangements for handling and holding in place complete dies
    • B21D37/145—Die storage magazines
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
    • B21D—WORKING OR PROCESSING OF SHEET METAL OR METAL TUBES, RODS OR PROFILES WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
    • B21D5/00—Bending sheet metal along straight lines, e.g. to form simple curves
    • B21D5/02—Bending sheet metal along straight lines, e.g. to form simple curves on press brakes without making use of clamping means
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J9/00—Programme-controlled manipulators
    • B25J9/16—Programme controls
    • B25J9/1656—Programme controls characterised by programming, planning systems for manipulators
    • B25J9/1664—Programme controls characterised by programming, planning systems for manipulators characterised by motion, path, trajectory planning
    • B25J9/1666—Avoiding collision or forbidden zones
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/18—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form
    • G05B19/406—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form characterised by monitoring or safety
    • G05B19/4069—Simulating machining process on screen
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/18—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form
    • G05B19/4097—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form characterised by using design data to control NC machines, e.g. CAD/CAM
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM]
    • G05B19/41815—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM] characterised by the cooperation between machine tools, manipulators and conveyor or other workpiece supply system, workcell
    • G05B19/41825—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM] characterised by the cooperation between machine tools, manipulators and conveyor or other workpiece supply system, workcell machine tools and manipulators only, machining centre
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/31—From computer integrated manufacturing till monitoring
    • G05B2219/31352—Expert system integrates knowledges to control workshop
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/33—Director till display
    • G05B2219/33002—Artificial intelligence AI, expert, knowledge, rule based system KBS
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/35—Nc in input of data, input till input file format
    • G05B2219/35189—Manufacturing function, derive gripper position on workpiece from cad data
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/35—Nc in input of data, input till input file format
    • G05B2219/35192—From design derive sequence of bending so that bending is possible
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/36—Nc in input of data, input key till input tape
    • G05B2219/36268—From blank and finished entered shape, derive machining features
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/36—Nc in input of data, input key till input tape
    • G05B2219/36304—Divide into several machining processes, divide each also in several sub processes
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/39—Robotics, robotics to robotics hand
    • G05B2219/39105—Manipulator cooperates with moving machine, like press brake
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/39—Robotics, robotics to robotics hand
    • G05B2219/39467—Select hand as function of geometric form of hand
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/40—Robotics, robotics mapping to robotics vision
    • G05B2219/40054—Supply sheet to bending machine
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/40—Robotics, robotics mapping to robotics vision
    • G05B2219/40501—Using sub goal method of options for semi optimal path planning
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/45—Nc applications
    • G05B2219/45143—Press-brake, bending machine
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/49—Nc machine tool, till multiple
    • G05B2219/49068—Minimum cost adaptive
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Definitions

  • the present invention relates in general to methods and subsystems which may be provided in an intelligent bent sheet metal designing, planning and manufacturing system and the like, in particular, the present invention relates to a bending apparatus comprising a control means being configured to generate a plan which comprises a sequence of operations to be performed by the bending apparatus for bending workpieces comprising sheets of malleable material, said sequence of operations comprising a set of N bends for forming a finished workpiece from a stock sheet of malleable material, said control means comprising: proposing means for proposing, for an m th operation within the sequence of operations, a plurality of proposed operations including a plurality of proposed bends to be performed by said bending apparatus; subplan means for providing a proposed subplan that accompanies each proposed bend, and generating means for generating a plan including a sequence of bends from a first bend through an N th bend, by choosing each bend in the sequence of operations based upon the proposed bends and the proposed subplan that accompanies each proposed bend.
  • Such a control means is e.g. known from US-A-5 307 282.
  • Figs. 1-3 illustrate, in a simplified view, an example conventional bending workstation 10 for bending a sheet metal part (workpiece) 16 under the control of a manually created program downloaded to various control devices provided within the workstation.
  • the illustrated bending workstation 10 is a
  • Fig. 1 shows an overall simplified view of bending workstation 10.
  • Fig. 2 shows a partial view of a press brake 29, positioned to perform a bend on a workpiece 16.
  • the elements shown in Fig. 2 include a robot arm 12 having a robot arm gripper 14 grasping a workpiece 16, a punch 18 being held by a punch holder 20, and a die 19 which is placed on a die rail 22.
  • a backgage mechanism 24 is illustrated to the left of punch 18 and die 19.
  • bending workstation 10 includes four significant mechanical components: a press brake 29 for bending workpiece 16; a five degree-of-freedom (5 DOF) robotic manipulator (robot) 12 for handling and positioning workpiece 16 within press brake 29; a material loader/unloader (L/UL) 30 for loading and positioning a blank workpiece at a location for robot 12 to grab, and for unloading finished workpieces; and a repositioning gripper 32 for holding workpiece 16 while robot 12 changes its grasp.
  • 5 DOF five degree-of-freedom
  • robot 12 for handling and positioning workpiece 16 within press brake 29
  • L/UL material loader/unloader
  • L/UL material loader/unloader
  • a repositioning gripper 32 for holding workpiece 16 while robot 12 changes its grasp.
  • Press brake 29 includes several components as illustrated in Figs. 1-3. Viewing Fig. 3, press brake 29 includes at least one die 19 which is placed on a die rail 22, and at least one corresponding punch tool 18 which is held by a punch tool holder 20. Press brake 29 further includes a backgage mechanism 24.
  • robot arm 12 includes a robot arm gripper 14 which is used to grasp workpiece 16.
  • material loader/unloader 30 includes several suction cups 31 which create an upwardly directed suction force for lifting a sheet metal workpiece 16, thereby allowing L/UL 30 to pass workpiece 16 to gripper 14 of robot 12, and to subsequently retrieve a finished workpiece 16 from gripper 14 and unload the finished workpiece.
  • loader/unloader (L/UL) 30 will lift a blank workpiece 16 from a receptacle (not shown), and will raise and move workpiece 16 to a position to be grabbed by gripper 14 of robot 12.
  • Robot 12 then maneuvers itself to a position corresponding to a particular bending stage located within bending workstation 10.
  • stage 1 comprises the stage at the leftmost portion of press brake 29, and stage 2 is located to the right of stage 1 along die rail 22.
  • robot arm 12 will reposition its robot arm gripper 14 to hold workpiece 16. Once gripper 14 is holding workpiece 16, die 19 will be disengaged by releasing press brake 29. Robot 12 then maneuvers and repositions workpiece 16 in order to perform the next bend in the particular bend sequence that has been programmed for workpiece 16. The next bend within the bend sequence may be performed either at the same stage, or at a different stage, such as stage 2, depending upon the type of bends to be performed, and the tooling provided within press brake 29.
  • Repositioning gripper 32 shown in Fig. 1, is provided for this purpose.
  • workpiece 16 will be moved by robot 12 to repositioning gripper 32.
  • Repositioning gripper 32 will then grasp workpiece 16 so that robot gripper 14 can regrip workpiece 16 at a location appropriate for the next bend or sequence of bends.
  • the bending workstation 10 illustrated in Fig. 1 is controlled by several control devices which are housed separately, including an MM20-CAPS interface 40, a press brake controller 42, a robot controller 44, and a load/unload unit controller 46.
  • Press brake controller 42 comprises an NC9R press brake controller
  • robot controller 44 comprises a 25B robot controller, which are each supplied by Amada.
  • Each of press brake controller 42 and robot controller 44 have their own CPU and programming environments.
  • Load/unload unit controller 46 comprises a stand alone Programmable Logic Controller (PLC), and is wired to respective consoles provided for press brake controller 42 and robot controller 44.
  • PLC Programmable Logic Controller
  • controllers 42, 44, and 46 has a different style bus, architecture, and manufacturer. They are coordinated primarily by parallel I/O signals. Serial interfaces are provided for transporting bending and robot programs to the controllers, each of which is programmed in a different manner. For example, logic diagrams are used to program the PLC of the load/unload controller 46, and RML is used to program robot controller 44.
  • the overall design/manufacture process for bending sheet metal includes several steps. First, a part to be produced is typically designed using an appropriate CAD system. Then, a plan is generated which defines the tooling to be used and a sequence of bends to be performed. Once the needed tooling is determined, an operator will begin to set up the bending workstation. After the workstation is set up, the plan is executed, i.e., a workpiece is loaded and operation of the bending workstation is controlled to execute the complete sequence of bends on a blank sheet metal workpiece. The results of the initial runs of the bending workstation are then fed back to the design step, where appropriate modifications may be made in the design of the part in view of the actual operation of the system.
  • a plan is developed for bending workstation 10 in order to configure the system to perform a sequence of bending operations. Needed hardware must be selected, including appropriate dies, punch tools, grippers, and so on. In addition, the bending sequence must be determined, which includes the ordering and selection of bends to be performed by bending workstation 10. In selecting the hardware, and in determining the bending sequence, along with other parameters, software will be generated to operate bending workstation 10, so that bending workstation 10 can automatically perform various operations of the bending process.
  • a plan for a BM100 bending workstation includes generated software such as an NC9R press brake program and a 25B RML robot program.
  • Each of these programs may be created with the use of an initial part design created from a CAD system. Both the robot program and the bending program must be developed manually, and are quite labor-intensive.
  • Previously developed programs are classified by the number of bends and/or by the directions of the bends. Engineers examine each part style to determine if previously developed and classified programs may be used or whether a new program must be written. However, since each classified program typically supports only a narrow range of acceptable part dimensions, new programs must frequently be written by the engineers.
  • the final RML robot program when complete, is compiled and downloaded by the MM20-CAPS system 40 to robot controller 44.
  • the bending program is entered and debugged on a control pendant provided on press brake controller 42.
  • an operator After entering the robot and bending programs into the system, an operator performs several manual operations to walk the system through the several operations to be performed. For example, an operator will manually operate a hand-held pendant of the robot controller to manually move the robot to the loading and unloading positions, after which the interface console 40 will store the appropriate locations into the final RML program to be compiled and downloaded to robot controller 44.
  • the operator may control the system to follow the planned bend sequence, in order to determine the values for the backgage position (L axis) and the die rail position (D axis).
  • the Paper discusses several components of an overall intelligent manufacturing workstation, including features such as open architecture, the use of software modules that communicate via a query-based language, part design, operations planning, workstation control, and geometric modeling.
  • an effective intelligent manufacturing workstation should have open software, open controller and open mechanism architecture. That is, a machine tool user operating such a workstation should be able to add onto the software, the controller, and the mechanism architectures of the workstation in order to improve their functions.
  • the proposed architecture includes a designer 50, a bend sequence planner 52, a module 54 for sequence planning, execution and error handling, a modeler 56, a module 58 for sensor interpretation, and modules 60, 62 for process control and holding, and fixturing.
  • Each of the modules for sensor interpretation 58, process control 60, and holding and fixturing 62 are coupled to external machine and sensor drives 64.
  • a control subsystem 68 is formed by several of the modules, including sequence planning, execution and error handling module 54, modeler 56, and the modules for sensor interpretation 58, process control 60 and holding and fixturing 62.
  • Control subsystem 68 is shown as being implemented within a Chimera operating system. All of the modules may be connected to other factory systems 66, including, e.g., systems for scheduling, operations, and process planning.
  • a planner typically then produces a plan which will later be used to execute the manufacturing process.
  • the plan includes several instructions regarding the sequencing of machine operations to produce the desired part.
  • An optimal plan will result in a reduction of setup time, a reduction in the existence of scrap after production of the parts, an increase in part quality, and an increase in production rate.
  • the above-noted Paper recommends that as much specific knowledge as possible be separated from the planner so that the planner can be easily adapted to different machines and processes.
  • a "query-based" planning system is thus proposed which shifts the emphasis of the planner to asking expert questions, rather than attempting to act as a self-contained expert.
  • the above-noted Paper proposes that the controller use an off-the-shelf engineering UNIX workstation as the core computing resource.
  • the workstation may include in its back-plane an extension rack of special-purpose boards and an additional CPU that runs with a real-time version of the UNIX operating system, called CHIMERA-II. See, e.g., STEWART et al., Robotics Institute Technical Report, entitled “CHIMERA II: A Real-Time UNIX-Compatible Multiprocessor Operating System for Sensor Based Control Applications," Carnegie Mellon, CMU-RI-TR-89-24 (1989).
  • Geometric modeling is an important component in intelligent machining workstations.
  • Several modelers have been experimented with during a project in the Robotics Institute at Carnegie Mellon University.
  • a geometric modeler called “NOODLES” has been proposed for use as a modeler in an intelligent manufacturing workstation.
  • the NOODLES modeler is discussed by GURSOZ et al., in "Boolean Set operations on non-manifold boundary representation objects," in Computer Aided Design, Butterworth-Heinenmann LTD., Vol. 23, No. 1, January, 1991.
  • NOODLES system makes far fewer assumptions about what constitutes valid edge topologies, and thus overcomes problems with other modeling systems, which would enter into infinite loops when the edge topology of a geometric model would violate system assumptions.
  • bending apparatus/bending workstation a workstation or apparatus for performing modern sheet metal working functions, including bend operations.
  • bending sheets of malleable material - working of sheets of malleable material, such as sheet metal, including, and not limited to, up-action air bending, V bending, R bending, hemming, seaming, coining, bottoming, forming, wiping, folding type bending, custom bending, and so on.
  • operations plan - a sequence of operations to be performed by a part forming apparatus in order to form a finished part from a piece of unfinished material.
  • an operations plan comprises a sequence of operations to be performed by a bending apparatus for bending workpieces comprising sheets of malleable material, the sequence of operations including a bend sequence which includes all of the bends needed to form a finished bent workpiece.
  • subplan - a portion of a complete operations plan.
  • a subplan comprises a part of the information needed to set up and/or control a bending workstation/apparatus.
  • a bending apparatus comprising a control means being configured to generate a plan, that allows evaluation of a bending operation more effectively.
  • a bending apparatus comprising a control means being configured to generate a plan which comprises a sequence of operations to be performed by the bending apparatus for bending workpieces comprising sheets of malleable material, said sequence of operations comprising a set of N bends for forming a finished workpiece from a stock sheet of malleable material, said control means comprising:
  • an estimating device for estimating a cost to be associated with each proposed bend
  • the generating mechanism may generate a plan including a sequence of bends from a first through an Nth bend, by choosing each bend in the sequence of operations based upon the proposed bend, the proposed subplan that accompanies each proposed bend, and the estimated costs associated with each proposed bend.
  • the estimated costs associated with an nth bend in the sequence of N bends may comprise a k cost calculated based upon an estimated amount of time it will take the bending apparatus to complete one or more operations of the bend.
  • the estimated costs associated with an nth bend in a sequence of N bends may comprise an h cost calculated based upon an estimated total amount of time it will take the bending apparatus to complete one or more operations of each of the rest of the bends in the bend sequence that follow the nth bend.
  • the one or more operations of the bend which will be timed in order to calculate the k and h costs may comprise moving the workpiece from a tooling stage location of a preceding bend to a tooling stage location of the given bend.
  • the one or more operations of a given bend may also comprise installing, when setting up the bending apparatus, an additional tooling stage needed to perform the given bend.
  • the one or more operations of a given bend may also comprise repositioning of the gripper's hold on the workpiece before performing the given bend.
  • the proposing mechanism and the generating mechanism collectively comprise a bend sequence planning module
  • the subplan mechanism and the estimating mechanism collectively comprise a plurality of expert modules.
  • the expert modules may each operate the subplan mechanism and the estimating mechanism when the proposing mechanism proposes a proposed operation for performance as the mth operation within the sequence of operations.
  • the plurality of expert modules may comprise a holding expert module which is capable of operating the subplan mechanism to provide a proposed subplan, including information regarding a location on the workpiece at which the gripper can hold the workpiece while performing the bends of the bend sequence.
  • the plurality of expert modules may comprise a holding expert module which is capable of operating the estimating mechanism to estimate a holding cost, calculated based upon whether a gripper's hold on the workpiece is to be repositioned before performing a given bend.
  • the plurality of expert modules may comprise a tooling expert module which is capable of operating the subplan mechanism to provide a proposed tooling subplan that includes information regarding a position along a tooling stage at which the workpiece will be loaded into the bending apparatus in order to perform a given bend.
  • the tooling expert may also be capable of operating the estimating mechanism to estimate a cost based upon an amount of time to install, when setting up the bending apparatus, an additional tooling stage needed to perform a given bend.
  • the motion expert module may also be capable of operating the estimating mechanism to estimate a cost based upon a calculated travel time for moving the workpiece from a tooling stage location of one bend to a tooling stage location of a next bend.
  • the bend sequence planning module may be capable of querying each of the expert modules for a subplan and estimated costs.
  • each of the expert modules may be capable of responding to a query by returning a savelist to the bend sequence planning module, whereby the savelist includes a list of names of attributes, and values respectively corresponding to the attributes, to be saved by the bend sequence planning module.
  • control means includes a prioritizing mechanism for prioritizing proposed bends in accordance with bend heuristics determined based upon the geometry of the workpiece.
  • the generating mechanism may generate a plan, including a sequence of bends from a first through an Mth bend, by choosing each bend in the sequence of operations based upon the prioritized proposed bends and the proposed subplan that accompanies each proposed bend.
  • the prioritizing mechanism may be provided with a mechanism for discounting an estimated cost of a bend having a high priority and increasing an estimated cost for a bend having a low priority.
  • a determining mechanism may be provided for determining the time needed for, and the feasibility of, producing one or more parts with the bending apparatus based upon the generated plan.
  • the system may be provided with a mechanism for performing calculations of the costs of producing a given batch of parts, based upon the time determined by the determining mechanism.
  • the system may be provided with a mechanism for redesigning the part based upon the time and the feasibility determinations made by the determining mechanism.
  • the system may be further provided with a mechanism for scheduling manufacturing with the bending apparatus depending upon the determined amount of time for producing one or more parts.
  • control means may utilize a computerized method for selecting a gripper for holding a workpiece.
  • the gripper is selected for use in a bending apparatus for bending unfinished workpieces comprising sheets of malleable material.
  • the method includes reading information describing the geometry of a library of grippers to be chosen from, forming a set of available grippers excluding grippers that have certain undesired geometric features, and choosing a gripper from a set of available grippers.
  • the gripper is chosen as a function of the width of the gripper, the length of the gripper, and the knuckle height of the gripper.
  • the gripper may include a gripper for holding the workpiece while loading and unloading the workpiece into and from a die space of the bending apparatus.
  • the method may include a step of predicting, for each gripper within the set of available grippers, a repo number equal to an estimated number of times the bending apparatus will need to change the position at which the gripper is holding the workpiece in order to perform a complete sequence of bending operations on the workpiece.
  • the smallest predicted repo number is then determined, and the set of available grippers is adjusted to include the available grippers having a repo number equal to the smallest predicted repo number, before choosing (from among the set of available grippers) a gripper as a function of the gripper's width, length, and knuckle height.
  • the gripper may alternatively comprise a repo gripper for holding the workpiece while a robot changes its grip on the workpiece.
  • the method may be further provided with a step of constructing data representations of the respective intermediate shapes of the workpiece when repo operations are to be performed by the bending apparatus, and utilizing the intermediate shapes to determine which grippers are excluded from the set of available grippers. The grippers that cannot securely grasp the workpiece, considering all of the constructed intermediate shape representations, are excluded from the set of available grippers.
  • control means may comprise a computerized method for determining a location at which a gripper can hold a malleable sheet workpiece while a bending apparatus performs an mth operation on the workpiece.
  • the bending apparatus performs a sequence of operations, including the mth operation, in accordance with a bending plan.
  • the sequence of operations includes a sequence of bends from a first bend through an Nth bend, and the shape of the workpiece changes to several intermediate shapes as the bending apparatus progresses through the sequence of bends.
  • a set of topographic representations is formed by repeatedly generating, along edges of the workpiece, as a variable i is varied, a graphic representation of areas on the workpiece within which the gripper location can be without hindering performance of an ith operation.
  • a determination is made as to whether or not the performance of the ith operation will be hindered by taking into consideration the intermediate shape of the workpiece when the ith operation is performed.
  • the method further includes the step of determining the intersection of all the geographic representations within the set to thereby determine the areas common to the given plurality of operations in the sequence of operations.
  • the mth operation may include changing a robot's grip on the workpiece between bends in the sequence of bends, and/or performing a bend within the sequence of bends.
  • control means may utilize a computerized method for selecting tooling to be used in a bending apparatus for bending a workpiece comprising a sheet of malleable material.
  • the tooling includes at least a die and a punch, and the bending apparatus performs, utilizing the selected tooling, a sequence of operations comprising a sequence of bends from a first bend through an Nth bend.
  • the method comprises steps of reading information describing in the geometry of dies and punches, and forming sets of feasible dies and punches excluding dies and punches that have an insufficient force capacity to bend the workpiece and that are incapable of forming desired bends in the workpiece resulting in desired angles and desired inside radii.
  • the method includes a step of choosing an appropriate die and appropriate punch that most closely satisfies force, bend angle, and inside radii requirements, excluding punches that will likely collide with the workpiece as determined by failure of a geometric collision test.
  • the geometric collision test may be performed by modeling a finished 3D workpiece and, for each bend in the sequence of bends, aligning the modeled finished 3D workpiece between a model of each feasible punch and a model of a chosen die.
  • the bending apparatus is adapted to bend workpieces comprising sheets of malleable material, by performing a sequence of operations comprising a sequence of bends from a first bend through an Nth bend.
  • the method includes a step of deciding on an arrangement of a plurality of stages along the die rail and calculating lateral limits based upon the amount by which the workpiece extends beyond a side edge of a tooling stage for the bends of the sequence of bends.
  • the method includes determining a largest lateral limit for each side of the stage, and spacing adjacently arranged stages to have a gap between adjacent side edges that is greater than or equal to the larger of the determined largest lateral limits of the adjacent side edges.
  • a control means for generating a plan and for controlling a bending apparatus.
  • the plan comprises a sequence of operations to be performed by the bending apparatus, and the bending apparatus is adapted to bend workpieces comprising sheets of malleable material.
  • the sequence of operations includes a sequence of bends, from a first through an Nth bend, for forming a finished workpiece from a stock sheet of malleable material.
  • the system includes a setup planning mechanism for generating the sequence of bends and a setup subplan that includes information regarding the manner in which the bending apparatus is to be set up before commencing the first bend in the sequence of bends.
  • the system includes a forwarding mechanism for forwarding the setup subplan, once generated, to a signalling device for signalling commencement of setup operations to be performed in accordance with the setup subplan.
  • a finalize mechanism is further provided for generating detailed subplan information to complete the plan after the setup subplan has been generated. At least part of the detailed subplan information is generated after the commencement of setup operations has been signalled by the signalling device.
  • the setup subplan may include one or more of the following types of information: information regarding the layout of tooling stages; information regarding tooling die and punch profiles to be utilized in the bending apparatus; positions of tooling stages along a die rail of the bending apparatus; information regarding what type of gripper to use for manipulating the workpiece through the bend sequence; and information regarding what type of repo gripper to use for holding the workpiece while a gripper changes its grasp on the workpiece in between bends of the bend sequence.
  • the forwarding device may include a device for forwarding instructions to a sequencer module which directs performance of automated setup operations on the bending apparatus.
  • the forwarding device may also, or in the alternative, create a visual representation of setup operations to be performed on the bending apparatus so that a human operator can thereby perform the setup operations.
  • an embodiment of the present invention may be directed to a control means for performing setup operations on the bending apparatus so that the bending apparatus can be utilized to perform bending operations on workpieces comprising sheets of malleable material.
  • the bending apparatus includes a die, a tool punch holding mechanism, and one or more tooling stages. Each tooling stage includes a die mounted on the die rail and a tool punch held by the punch holding mechanism.
  • the system further includes a mechanism for receiving information regarding a location of each of the one or more tooling stages along the die rail, and a control mechanism for controlling a position of a guide member along at least one of a die rail and the tool punch holding mechanism based upon the received information so that at least one of the die and the tool punch can be aligned with reference to the guide member and so that the resulting tooling stage will be at a desired location along the die rail.
  • the control mechanism may be capable of positioning the guide member to be at a specified position along the die rail and to be within a certain distance from the die rail, whereby a die of a tooling stage to be aligned can be abutted against the guide member in order to properly position the tooling stage along the die rail.
  • the guide member may include a backgage finger of a mechanism for performing backgaging when loading a workpiece into the bending apparatus.
  • a control means for executing a plan for controlling a bending apparatus for bending workpieces comprising sheets of malleable material.
  • the plan includes a sequence of operations to be performed by the bending apparatus.
  • a sensor-based control mechanism is provided for performing an operation, including moving a workpiece from one position to another, with the bending apparatus utilizing a sensor output to modify the movements of the workpiece.
  • a measuring device measures an amount by which the movement of the workpiece was modified due to the sensor output, and a learned control mechanism performs the operation, including moving the workpiece from one position to another, without modifying the movement of the workpiece utilizing a sensor output.
  • the learned control mechanism controls performance of the operation based upon the amount measured by the measuring device.
  • the inventive system for generating a plan may provide an intelligent bending workstation environment/system which may be easily upgraded and integrated with additional or alternate hardware and software modules. Accordingly such a system can be used to economically produce very small batch sizes (of one or more workpieces) with high quality, and in a short amount of time.
  • a system is flexible and is able to accommodate new and different part styles in the design and manufacture process.
  • the system of the present invention is able to operate efficiently in large volume production, and to learn from initial production runs in order to maximize efficiency.
  • an additional feature of such a bending apparatus is that it is able to maintain quality of the produced parts throughout the process, and to avoid errors and collisions during execution of the process by the bending workstation.
  • the system of the present invention can provide an intelligent sheet metal bending workstation which makes small batches of sheet metal parts from CAD descriptions.
  • a process planner is provided that selects the necessary hardware (e.g., dies, punches, grippers, sensors) to be utilized by the bending workstation, determines bending sequences, and generates the necessary software to operate the bending machine.
  • control means can produce a plan for bending a sheet metal workpiece, in which the smallest number of tooling stages will be utilized to make the part. Accordingly, the system will efficiently and automatically produce the plan to be utilized by the bending workstation, set up the workstation and execute the plan.
  • the present invention therefore, comprises several systems, methods and sub-components provided in connection with a system for generating a plan which comprises a sequence of operations to be performed by a bending apparatus for bending workpieces comprising sheets of malleable material.
  • the bending apparatus has a gripper for gripping a workpiece while performing a bend, and the sequence of operation includes a set of N bends for forming a finished workpiece from a stock sheet of malleable material.
  • the system includes a proposing mechanism for proposing, for an mth operation within the sequence of operations, a plurality of proposed operations including a plurality of proposed bends to be performed by the apparatus.
  • the system includes a subplan mechanism for providing a proposed subplan that accompanies each proposed bend, and a generating mechanism for generating a plan including a sequence of bends from a first bend through an Nth bend, by choosing each bend in the sequence of operations based upon the proposed bends and the proposed subplan that accompanies each proposed bend.
  • the proposing mechanism may be designed so that it proposes bends among the complete set of N bends that are still remaining, or proposes bends among the complete set of bends that are still remaining less bends blocked due to constraints.
  • the proposing mechanism may propose, for an mth operation, a repositioning of a gripper's hold on the workpiece.
  • the generated plan further includes at least part of the proposed subplans that accompany the chosen bends.
  • the system may further include a mechanism for representing the mth operation as an mth level of a search tree.
  • the proposed subplans may include setup and control information for the bending apparatus, and may further comprise final locations on the workpiece at which the gripper will grip the workpiece while performing the bends of the bend sequence.
  • the proposed subplans may further include ranges of locations on the workpiece at which the gripper can grip the workpiece while performing the bends of the bend sequence.
  • the proposed subplans may comprise: numbers representing a predicted number of repositionings of the gripper needed to complete the sequence of bends, indications that the next bend in the sequence cannot be performed unless the gripper is first repositioned, and/or locations on the workpiece at which a repositioning gripper (i.e., a repo gripper) will grip the workpiece while performing a repositioning operation.
  • the proposed subplans may include: tooling stages to be utilized to perform the bends in the bend sequence, positions along a tooling stage at which the workpiece will be loaded into the bending apparatus in order to perform the bends, and/or motion plans for maneuvering around tooling stages in performing the bends.
  • the present invention is further exemplified by a plurality of listings which are provided in the Appendices, wherein:
  • Fig. 5 illustrates a block diagram of an embodiment of a planning and control system 70 for an intelligent manufacturing bending workstation.
  • planning and control system 70 includes a CAD system 74, a bend sequence planner 72, a plurality of experts (sub-planners), and a sequencer 76.
  • Planning and Control System 70 is connected to hardware and sensors 78 via an interface 77.
  • the experts include a tooling expert 80, a holding expert 82 and a motion expert 84. Additional experts may be provided, such as sensing expert 85 illustrated in dotted lines.
  • Bend sequence planner 72, experts 80, 82, and 84, and CAD system 74 may be implemented within a UNIX-compatible environment on a workstation computer such as a Sparc 10 Sun OS v.4.1.3.
  • Sequencer 76 may be implemented within an additional CPU coupled to the Sun workstation via a bus adaptor.
  • the bus adaptor may comprise a BIT 3 VME-to-VME bus adaptor which extends between the Sun workstation and a remote VME bus passive back-plane.
  • the passive back-plane may hold several interface mechanisms such as VME (Virtual Memory Extension) boards, which together form part of interface 77 as illustrated in Fig. 5.
  • Sequencer 76 may be implemented within a real-time UNIX-compatible multiprocessor operating system such as CHIMERA, and may be run by the additional CPU which is provided in the computer workstation's back-plane. Accordingly, in the illustrated embodiment (shown in Fig. 5), CAD system 74, bend sequence planner 72, experts 80, 82, 84 (and 85) and sequencer 76 are each implemented primarily with software which controls the operations of a computer utilizing a UNIX-compatible operating system. Sequencer 76 is implemented within a real-time UNIX-compatible multiprocessor operating system such as CHIMERA.
  • CAD system 74 is utilized to design a sheet metal configuration, by defining the shape of a stock (flat) sheet metal part and the bends to be performed on the stock part to form a desired three-dimensional finished part.
  • CAD system 74 forms one or more information files which describe the part.
  • the CAD system maintains in memory, and visually, a three-dimensional representation of the sheet metal part in parallel with a two-dimensional representation of the part. The designer may modify the design by adding or removing details to or from either representation.
  • CAD system 74 may also perform functions such as gathering and/or generating information needed for geometric modeling and requesting advice from bend sequence planner 72 as to whether certain design features can be implemented by the bending workstation.
  • Bend sequence planner 72 operates in cooperation with tooling expert 80, holding expert 82, motion expert 84, and any other experts (e.g., sensing expert 85) to produce a plan for complete part production by a bending workstation of the part designed with the use of CAD system 74. Bend sequence planner 72 performs functions such as proposing a particular bend in a hypothetical bend sequence, and determining what initial steps must be performed by the system in order to execute such a bend having a position within the hypothetical bend sequence.
  • bend sequence planner 72 may query tooling expert 80 as to what tooling would be needed to execute the proposed bend, querying holding expert 82 as to how the workpiece can be held while performing the proposed bend, and querying the motion expert 84 as to whether and to what extent the robot (which is holding the workpiece) can be manipulated to assist in making the bend. If a sensing expert 85 is provided, bend sequence planner 72 might query sensing expert 85 as to whether a particular sensor-based control strategy is needed in order to facilitate the execution of the proposed bend by the workstation and the costs associated with a particular sensor-based control strategy.
  • Bend sequence planner 72 may be configured to continually propose bends from a first bend consecutively to a last bend in a complete bend sequence, thus resulting in a complete set of bends to perform the final workpiece. Once the successful final bend sequence has been generated in this manner, bend sequence planner 72 may be configured to generate a final plan (which includes a general list of steps and accompanying information needed to control execution of the various hardware elements of the workstation), and forward the plan to sequencer 76.
  • a final plan which includes a general list of steps and accompanying information needed to control execution of the various hardware elements of the workstation
  • Sequencer 76 directs execution of the plan developed by bend sequence planner 72. Sequencer 76 interprets commands given by bend sequence planner 72 in the resulting plan, and controls timing of the various commands by parsing the commands and information accompanying the commands and placing them into queues provided for each of the main hardware elements of the sheet metal bending workstation.
  • Controller 75 comprises a plurality of tasks which correspond to the various hardware elements of the workstation. Each task is activated by the sequencer in an appropriate manner in accordance with the plan forwarded by the planner.
  • Bend sequence planner 72 and the several sub-planners including, e.g., tooling expert 80, holding expert 82 and motion expert 84, (and sensing expert 85), form a planning system 71.
  • Bend sequence planner 72 analyzes the designed part (Sheet metal workpiece), provided by CAD system 74, and offers a bend sequence to be performed by the bending workstation. Planner 72 utilizes a state-space search method in order to determine an efficient sequence of bend operations that can be utilized by the bending workstation. Planner 72 converses with tooling expert 80, holding expert 82 and motion expert 84 in order to obtain the information it needs to make its decisions.
  • Tooling expert 80 responds to queries made by planner 72, and provides information to the bend sequence planner such as which tools will be needed for a particular bend operation or bend sequence. In addition, tooling expert 80 may inform bend sequence planner 72 of the arrangement of tools within the workstation. Tooling expert, in conjunction with planner 72, will attempt to design a setup of tooling so that the fewest number of stages/toolings are utilized to make a particular part, i.e., to execute a complete bend sequence for making the part.
  • Holding expert 82 makes holding-related determinations, such as, e.g., whether the robot can hold the workpiece while a particular bend, specified by bend sequence planner 72, is being performed. Holding expert 82 may also determine the location at which the robot should hold the workpiece so that the workpiece may be maneuvered through a series of bends, without collision, and without the need to change the robot's grasp on the workpiece. In addition, holding expert 82 may determine the position at which the repositioning gripper should hold the workpiece when the robot's grasp is being changed, and where suction cups 31 of loader/unloader (L/UL) 30 should be placed during unloading and loading of the workpiece.
  • L/UL loader/unloader
  • Motion expert 84 is responsible for generating a motion plan, i.e., the manner in which the robot should be maneuvered in order to move the workpiece through various spaces and along various routes as needed to execute the bends.
  • Bend sequence planner 72 and the respective experts may be modular to communicate with each other in a query-based manner. For example, before deciding to include a particular bend as part of the bend sequence, bend sequence planner 72 may query tooling expert 80 as to whether there are sufficient tools to handle the bend. Bend sequence planner 72 will then await a response from tooling expert 80. Tooling expert 80 will recognize the query from bend sequence planner 72, and will return with a response, e.g., indicating that there are sufficient tools to handle that particular bend noted by bend sequence planner 72. By way of example, bend sequence planner 72 may also ask holding expert 82 if robot arm gripper 14 can remain holding onto the workpiece during a particular bend operation without repositioning its grasp of the workpiece. Holding expert 82 will then respond to the query made by bend sequence planner 72, and bend sequence planner 72 will then utilize the information to perform its next determination.
  • Each of the modules of planning system 71 utilizes one or more functions provided by a geometric modeling library (not shown) in order to model the relative interactions and positions of each of the hardware components of the system as may be needed in making their determinations.
  • the setup process can be performed completely manually, or it may be automated in full or in part with the use of automated tool changers.
  • the manual activities to be performed during the setup process may include downloading program data to dedicated controllers such as those illustrated in Fig. 1.
  • each stage (stage 1 and stage 2 as illustrated in Fig. 5C) must be set up by placing a plurality of die segments 810a, 810b, and 810c in stage 1, and 811a, 811b, and 811c for stage 2 along die rail 22.
  • a human operator will typically measure the distance from the edge of the die rail 22 to a particular edge of the die corresponding to each stage. For example, a measurement may be made from the left edge of die rail 22 to the left edge of each die set for each stage in order to position the die segments corresponding to each stage.
  • a mechanism may be provided for automatically providing a guide that can be used by the setup operator to place the die segments at the appropriate location along die rail 22.
  • a mechanism may comprise a backgage finger 88 which can be automatically positioned at a particular edge of each stage along die rail 22.
  • backgage finger 88 may be first located at location A for purposes of abutting first die segment 810a against backgage finger 88, and subsequent installment of second and third die segments 810b and 810c. After aligning die segments for stage 1, backgage finger 88 may be automatically positioned to the next stage, i.e., stage 2. More particularly, backgage finger 88 may be positioned at one side of the die corresponding to stage 2.
  • backgage finger 88 is positioned at the left edge of die 811. While backgage finger 88 is at that position, first die segment 811a may be placed along die rail 22 and abutted against backgage finger 88 for alignment. Thereafter, die segments 811b and 811c may be placed on and secured to die rail 22.
  • Fig. 5B illustrates the main components for controlling the backgage finger 88 to assist in positioning an alignment of dies 810 and 811.
  • the subsystem comprises an input control module 87a which includes a mechanism for instructing backgage servo controller 87b to move backgage finger 88 to one or more particular stage locations.
  • alignment control module 87a may be provided in control portion 75 of planning and control system 70, while backgage servo controller 87b may be provided with an interface 77. More specifically, controller 75 may be provided with a backgage task module.
  • the backgage task module may be provided with a backgage finger die-alignment function which may be called by the backgage task module. In calling the die-alignment function, the backgage task module may activate and control a backgage servo controller through the use of a second level backgage device driver 206 (see Fig. 6), which in turn interacts with an appropriate level 1 device driver such as an I/O device driver 220 which interacts with a parallel I/O card connected to the backgage hardware of the bending workstation.
  • a second level backgage device driver 206 see Fig. 6
  • Another manual step that can be performed is positioning and/or adjusting of the punch holders 20.
  • standard steps may be performed to align tool punch segments so that they are properly seated within each punch holder 20 and correspond to the associated die segments. This may comprise operating the press so that the die segments and corresponding tool punch segments are compressed against each other with a set amount of force.
  • other standard adjustments and procedures known to those skilled in the art, may be performed during setup. For example, loader/unloader 30 may need to be adjusted so that suction cups 31 are properly positioned with respect to the workpiece 16.
  • Workstation 10 may be configured to be controlled automatically by the planning system, without any need for human intervention.
  • the planning system may be configured to download appropriate components of the plan to the appropriate control modules.
  • sequencer 76 is implemented within a real-time UNIX-compatible shell such as an Ironics IV-3230 computer with a CHIMERA II operating system. Additional information regarding possible implementations of a real-time scheduler such as sequencer 76 is provided in the CHIMERA manual by Stewart, Schmitz and Khosla, entitled "CHIMERA II Real-Time Programming Environment, Version 1.02" (October 29, 1990). Sequencer 76 schedules the general execution of the generated plan by control system 75, which utilizes interface architecture 77 to communicate with various hardware elements and sensors within the system, depicted as hardware and sensors 78 in Fig. 5.
  • Fig. 6 depicts in greater detail, sequencer 76, control system 75, and interface architecture 77.
  • sequencer 76 is connected to bend sequencer planner 72 and is further connected to a plurality of modules which comprise control system 75.
  • the modules of control system 75 include a robot task 92, a press and L/UL task 94, a backgage task 96, a motion library 98, a speed control module 102 and a collision detection module 100.
  • Interface architecture 77 comprises a set of level 2 device drivers and another set of level 1 device drivers.
  • the level 2 device drivers may include robot DD 202, press and L/UL DD 204, backgage DD 206, gripper DD 208, gripper sensor DD 210, droop sensor DD 212, backgage sensor DD 214, and angle sensor DD 216.
  • the level 1 device drivers may include respective device drivers 220, 222 and 224 for one or more parallel I/O VME cards, one or more A/D VME converter cards, and a robot servo control card.
  • the first level device drivers comprise a UNIX-like interface, with commands supported including open(), close(), read(), write(), ioctl(), and mnap() commands.
  • the first level device drivers standardize the interface to the I/O ports to which the hardware devices are attached, such as parallel I/O ports, analog/digital converters and a robot servo control mechanism.
  • the second level device drivers form an interface between the various modules of the control system 75 and the first level device driver.
  • the second level device drivers may be implemented with the use of a standard form as disclosed in the above-noted CHIMERA manual.
  • a software interface system may be provided which is reliable, portable, and has code which is easily readable. Specific details regarding the device drivers, and examples implementations thereof, are provided in the above-noted CHIMERA manual, which has been incorporated by reference herein.
  • VME cards which are the actual I/O ports connecting the computer to the hardware elements
  • such cards may include, as noted above, one or more parallel I/O cards, such cards preferably having optically isolated connections between the computer and the various hardware elements connected thereto.
  • the VME cards may include one or more Geonics motion two axis servo control cards II MCCII and one or more A/D converters having sufficient a number of channels and bit resolution, e.g., an A/D converter with 16 channels and 12 bit resolution, such as the IMV-1645 Ironics (Pentland-Burr-Brown MPV 950S).
  • the parallel I/O cards may include an 80-channel (with 64 usable channels) Xycom XVME-240 card and/or 32-channel digital output boards such as the Xycom XVME-220 and/or XVME-212 boards.
  • One or more A/D converters can be provided for inputting information such as reading various data produced by the sensors included in the workstation, such as a gripper sensor, droop sensor, backgage sensor, and/or angle sensor.
  • Each of the robot task 92, press and L/UL task 94, and backgage task 96 control the appropriate device drivers for controlling the corresponding hardware elements of the bending workstation.
  • Several functions which must be performed during execution of various motion-related functions may be provided in motion library 98. Such functions may include kinematics, trajectory calculations and filtering. Any control functions relating to speed control, i.e., controlling the speed with which various physical elements (such as the robot) of the bending workstation are moved, may be implemented within speed control module 102.
  • Collision detection module 100 is provided in order to perform collision detection which is needed in certain motion control processes during execution of the bend process.
  • Motion library 98 may further include dynamic motion control and sensor-based motion control modules which directly communicate with the second-level device drivers for dynamically controlling the movement of various components of the bending workstation and for changing such control in accordance with sensor-based signals produced by the various sensors provided in this system.
  • the computer be optically isolated from the actual hardware connections to prevent damage that may be caused by surges present at the hardware components.
  • Other reasons for optically isolating the parallel I/O cards is to protect the computer and the cards and to prevent the occurrence of ground loops.
  • the A/D converters be optically isolated from the sensors.
  • Bend sequence planner 72 of the embodiment shown in Fig. 5A performs three main functions. It generates a bend sequence, including accompanying operations associated with each bend, queries experts as to the consequences of the bend sequence as it is generated, and as to further plan details (subplans) needed to accomplish the generated bend sequence, and compiles all gathered/generated information in order to form an overall plan.
  • the plan specifies the steps needed to execute the bend sequence by a control system which controls operations of the sheet metal bending workstation.
  • Each of the experts of the illustrated planning system 71 performs three main functions when requested by planner 72. They each determine an incremental cost for performing an individual step within the bend sequence, develop proposed/intermediate plan information, and communicate the incremental cost and plan information to bend sequence planner 72.
  • the proposed/intermediate plan information includes two types of information: definite information and indefinite information.
  • holding expert 82 will know which regions of the workpiece may be grasped by the robot grasper to perform a given bend within a bend sequence (the grasp regions being definite), but will not yet know the exact grasp location (the precise grasp location being indefinite).
  • a temporary (indefinite) grasp location will be assigned by the holding expert 82, which can be verified at a later time.
  • sequence planner will query each expert as to the consequences of a bend sequence as it is generated.
  • the consequences of the bend sequence are represented in terms of cost.
  • the costs of the bend sequence as it is generated may be determined as a function of one or more of: the amount of time that it takes to perform a particular operation within the bend sequence, the extent to which an operation within the bend sequence will affect the accuracy of the operation and the quality of the resulting workpiece, whether or not there are any safety concerns associated with performing a particular operation at a particular point in a bend sequence, and whether there are any heuristics which, if taken into account, would suggest performing one operation instead of another at a particular point in the bend sequence.
  • Bend sequence planner 72 may query experts for information such as what tool profile should be utilized to perform certain bends of the bend sequence, what stage segments will be needed to form a given stage which will be needed to perform a bend, and where can/should the robot gripper grasp the workpiece in performing one or more bends of the bend sequence.
  • planner 72 may query the experts as to when a repositioning of the workpiece should be performed in the bend sequence, and how should the robot and the workpiece be moved in order to execute various operations throughout the sequence, such as a bend, repositioning, workstation load, and/or a workstation unload.
  • Fig. 7 represents, in a high level flow chart, the major steps performed by an example embodiment of bend sequence planner 72.
  • a first step S1 parallel design processing is performed by CAD system 74.
  • the parallel design processing may comprise, among other functions, labeling various geometries corresponding to respective portions of the workpiece, the resulting labels being used later (in step S3) by the bend sequence planner to determine whether heuristics should be considered in generating the bend sequence plan.
  • step S2 a heuristics framework is produced to guide the bend sequence planner in choosing the bends that will form the bend sequence.
  • a partial order of bending steps is computed that complies with certain specified heuristics.
  • step S3 a state-space search algorithm is performed which will be influenced by the heuristics framework.
  • the state-space search algorithm performs an analysis of the implications of performing various bends in a prescribed order, by assigning costs to each bend in step S4.
  • geometric reasoning is utilized, e.g., to determine the physical implications a particular bend will have by modeling the machine and the resulting workpiece as they relate to each other during the execution of each bend.
  • step S3 A particular sequence of bends is thus developed in step S3, which can be executed to produce the desired finished workpiece.
  • step S6 a determination is made in step S6 as to whether or not a complete plan, including a complete bend sequence, has been generated. If a plan cannot be formed for the design that has been specified, the process returns to step S1, where the workpiece may be redesigned to form a part design for which an operational plan can be created.
  • step S6 If a determination is made in step S6 that a complete plan was produced, the process will proceed to step S7, and the complete plan will be forwarded, using FEL, to the sequencer, or the plan may be stored in a file for later retrieval and execution by the sequencer.
  • the state-space search algorithm will preferably comprise an A* algorithm, such as disclosed, e.g., by Nils J. Nilsson in "Problem-Solving Methods in Artificial Intelligence” McGraw-Hill Book Company, 1971, pages 43-67.
  • the cost assignment step S4 may consider variables such as robot motion, gripping positions, the need for regripping, the need to change the gripper, tooling positions, and the need to change the tools. High costs are assigned for variables that will be time consuming, sacrifice quality, and/or expose the system to high risk.
  • the above-described operations planning method can be termed generative planning (since it automatically generates a bending plan), with weak heuristics and state-space searching.
  • a human inputs the design.
  • a heuristics framework is defined using heuristics which are called "weak heuristics" because they comprise only a limited set of rules. Possible bends are considered, and costs are assigned to each considered bend. The costs assigned to the bends are influenced by the heuristics framework by augmenting or discounting the cost of a particular bend. A sequence of bends of the least total cost is chosen utilizing a state-space searching algorithm.
  • Generative planning with weak heuristics as disclosed herein should be contrasted with other approaches to operations planning.
  • One such approach includes variant planning with case-based reasoning.
  • variant planning a human inputs a design of a new part, and the design is coded according to an index. The index is used to look up an old design which best resembles the current part to be designed and the problems to be solved.
  • a human operator edits the old plan to solve the new problems, e.g., by editing an RML program.
  • One of the problems noted with variant planning is that a similar design may require different or divergent solutions, which will not be discovered by comparison to old plans.
  • a generative planning system with strong heuristics lacks the flexibility and intelligence of a generative planning system with weak heuristics, and will likely be unable to handle unorthodox problems. Such a system has no understanding as to what heuristics work better in a particular situation, and which heuristics should be discarded. Moreover, such a system will be incapable of developing a plan in many cases.
  • Sheet metal bending heuristics can be taken into account by the bend sequence planner of the present invention.
  • Several exemplary bend heuristics will be described as follows.
  • One heuristic is to bend internal tabs early.
  • Fig. 25A illustrates a workpiece 16 having an internal tab 33 which is to be bent along bend line 34a.
  • the internal tab 33 be bent along bend line 34a first.
  • a workpiece 16 which includes outer bend lines 35a, 35b, 35c, and 35d, along with inner bend lines 36a, 36b, 36c, and 36d.
  • outer bends corresponding to outer bend lines 35a-35d be performed before the bends corresponding to inner bend lines 36a-36d.
  • Fig. 25C illustrates a workpiece having shorter bends along bend lines 37a and 37b, and longer bends along bend lines 38a and 38b. Accordingly, it is preferred that the bends along bend lines 37a and 37b be performed before the bends along bend lines 38a and 38b.
  • Fig. 25D illustrates a workpiece 16 having an outside face 39a and an inside face 39b which each abut each other at a corner 390. If the bend corresponding to the inside face was done first, then, when performing the bend corresponding to the outside face 39a, the press would not be able to cause the flange to be bent beyond its intended 90° angle. Accordingly, when the outside face springs back, it will not be flush with the end portion of inside face 39b.
  • a workpiece 16 is shown to include two tabs 26a and 26b, which are each to be bent along bend lines 27a, 27b, respectively. Since the bend lines 27a and 27b are co-linear, in accordance with the heuristic, it is preferred that the bends along those bend lines will be performed simultaneously.
  • heuristics are only examples of the types of heuristics which may be taken into account by the bend sequence planner of the present invention.
  • a larger or smaller set of heuristics, including all or a portion of the above-listed heuristics, may be utilized by the bend sequence planner.
  • a list of key features may be created which describe various geometric features of the workpiece which can then be utilized by the bend sequence planner in applying the heuristic rules.
  • a list of key features may be described with respect to the example workpiece 16 illustrated in Fig. 8. Several features may be deduced from workpiece 16, while it is still in its 2D state. An example of such features may include the flange number, the width of the flange, and the height of that flange.
  • values may be defined which specify an angle-class, i.e., a class of flanges which all have the same bend angle.
  • Additional features which may be labeled to avoid extra searching in the search space include an indication that the part that is symmetric around one or more axes.
  • Fig. 9 illustrates a workpiece 16 and a search tree 15 corresponding thereto.
  • Workpiece 16 has an axis of symmetry Y which is divided down the middle, running longitudinally through workpiece 16. Accordingly, at the first level of the search, the nodes corresponding to bends 3 and 5 have been eliminated (as indicated by the circles surrounding these bends) because they are symmetrical with nodes 2 and 4. There is no need to also evaluate and search through bends 3 and 5 at the first level, since the same effective results would be obtained if the search started with the bend corresponding to those nodes as opposed to either of bends 2 and 4. If the first bend chosen is bend 1, at the next level of the search, bends 2 and 4 are still symmetrical with bends 3 and 5.
  • bend-related operations which cannot be performed at certain points in the operations sequence being planned. These bend-related operations can be constrained to (or excluded from) certain locations in the bend sequence by using a mechanism referred to as a "constraint".
  • a feature extraction module may be provided to automatically label geometric features from geometric models produced by the design system (e.g., using data structures similar to those indicated above), and the geometric feature labels can be used to form legal phrases (called constraints) in an interface communication language, such as FEL.
  • Constraints may be defined by using a data structure that allows a particular arrangement of bend operations to be specified, in varying degrees of flexibility. For example, for a four-sided part 16 as illustrated in Fig. 26A, the following constraint statement can be used to specify the order in which bends 1, 2, 3, and 4 are performed:
  • the constraint statements may also include grouping operators, which require that certain bend operations be grouped together with no limitation on the order of the bend operations within the group. For example, the following constraint statement requires that bends 2 and 3 be before bend 4 in the bend sequence, and that bends 2 and 3 be grouped together with no bend operations therebetween:
  • constraint statement includes the above grouping constraint expression, as well as an additional constraint expression which further specifies that bend 1 must be before bend 4 without any additional limitations as to the inclusion and arrangement of the other operations with respect to bends 1 and 4:
  • Some additional example constraint expressions may include ( * 7 ), which means that bend operation 7 must be performed as the last bend operation in the sequence, and ( * 7 ? ), which means that bend operation 7 must be performed as the second to last bend operation in the sequence.
  • constraints may be expanded to include boolean operators such as NOT, OR, and AND.
  • boolean operators such as NOT, OR, and AND.
  • a constraint which uses a NOT operator could be ( * NOT 7), which would mean that the seventh bend operation could not be the last operation of the sequence.
  • constraints there is virtually no limit to the types of constraints that can be specified, and any entity in the planning system, including the various experts as well as a human operator of the bend sequence planner, can define constraints.
  • a constraint manager may be provided, e.g., within the bend sequence planner, in order to help maintain the consistency of constraints and resolve conflicts that arise between constraints.
  • the types of constraints may include constraints for (1) channels (e.g., as shown in Fig. 26B), (2) angle bends, where the bend line for the flange to be bent intersects and is close to a non-end point portion of a bend line of another bend (and both of the bends are to be performed in the same direction, e.g., they are both positive bends) (e.g., as shown in Fig. 27A), and (3) flanges which when bent form a corner with an outside flange and an inside abutting flange (e.g., as shown in Fig. 27C).
  • constraints for (1) channels e.g., as shown in Fig. 26B
  • angle bends where the bend line for the flange to be bent intersects and is close to a non-end point portion of a bend line of another bend (and both of the bends are to be performed in the same direction, e.g., they are both positive bends)
  • flanges which when bent form a corner with an outside
  • the constraint expression for the channel illustrated in Fig. 26B usually must be ( * 2 * 1 * 3 * ), even though a common heuristic prefers that bends on outer bend lines be performed before those of inner bend lines, which might suggest a constraint of ( * 3 * 2 * 1 * ). This conflict in constraint expressions, if it existed, would have to be resolved in favor of the channel constraint ( * 2 * 1 * 3 * ).
  • a human operator of the bend sequence planner may define a constraint expression which groups all bends on each side of a part together, so that less time will be spent by switching between sides of the part when performing a search for a solution bend sequence.
  • Fig. 27B shows a part with several bends on each side of the part, where it may be appropriate to group the bends for each side, e.g., by using the following constraint expression:
  • constraints may conflict, a mechanism should be provided for resolving conflicts.
  • a constraint manager may be provided within the bend sequence planner for this purpose.
  • a possible prioritization scheme could simply discard or ignore constraint expressions that have a higher assigned priority.
  • the priority assigned to constraint expressions could depend upon what type of constraint it is. For example, human input constraints could be assigned the highest priority, with machine constraints, part constraints, and optimization constraints being assigned respective lower priorities. Accordingly, machine constraints would have the second to highest priority, part constraints would have the third highest priority, and optimization constraints would have the fourth highest (i.e., the lowest) priority.
  • a human input constraint is a constraint input by a human operator controlling the bend sequence planner through a human interface.
  • a machine constraint is a constraint dictated by limitations of the machines and tooling (e.g., a channel constraint).
  • a part constraint is a constraint dictated by the features of the part (e.g., a constraint dictated by the presence of inside and outside abutting corners).
  • Optimization constraints are constraints that are created in order to speed up the search for a bend sequence (e.g., a constraint to group bends together that are on a particular side of the part).
  • an algorithm may be provided which first checks for the presence of common operations within a given pair of constraint expressions. If there is a common operation among the constraint expressions, they may then be merged together in order to determine if they conflict. For example, if (*1*2*) was merged with (*2*3*), the resulting merged constraint expression would be (*1*2*3*). If (*1*2*) was merged with a conflicting expression such as (*2*1*), a null would be the result, thereby indicating that the constraint expressions conflict with each other.
  • bends 5 and 6 in Fig. 8 it is preferred to have the bends performed simultaneously.
  • heuristics may be provided in order to influence the search performed by the bend sequence planner so that simultaneous bending of co-linear bends is preferred and thus more likely to become part of the bend sequence formed by the search.
  • constraints may be specified using constraint expressions to require that certain compatible co-linear bends be performed simultaneously if possible (i.e., if the constraint expression does not conflict with a constraint expression of higher priority).
  • a solution is obtained by applying operators to state-descriptions until an expression describing a goal state is obtained.
  • a start node is associated with an initial state-description, and successors of the start node are calculated using operators that are applicable to the state-description associated with the node. By calculating all of the successors of a node, the node is thereby expanded.
  • Pointers are set up from each successor node back to its parent node.
  • the pointers may later be used to indicate a solution path back to the start node, when a goal node is finally found.
  • the successor nodes are checked to see if they are goal nodes by checking the associated state-descriptions corresponding to the successor nodes to see if they describe the goal state. If a goal node has not yet been found, the process of expanding the nodes, and setting up corresponding pointers, continues. When a goal node is found, the pointers are traced back to the start node to produce a solution path. The state-description operators associated with the arcs of the path are then assembled into a solution sequence.
  • the above-described steps form a state-space search algorithm. Variations of the above-described algorithm may be defined by the order in which the nodes are to be expanded. If the nodes are expanded in an order in which they are generated, the search method is called a breadth-first method. If the most recently generated nodes are expanded first, the method is called a depth-first method. Breadth-first and depth-first methods are blind-search algorithms, since the order in which the nodes are expanded is unaffected by the location of the goal node.
  • Heuristic information about the overall nature of the graph and the general direction of the goal, can be utilized to modify the search process. Such information can be used to help direct the search toward the goal, in an attempt to expand the most promising nodes first.
  • One type of heuristic search method is described, e.g., by Nils J. Nilsson in "Problem-Solving Methods in Artificial Intelligence," noted previously.
  • Blind-search algorithms such as breadth-first or depth-first, are exhaustive in their approach to find a solution path to a goal node. In application, it is often impractical and time-consuming to use such methods, because the search will expand an excessive number of nodes before a solution path is found. Such an exhaustive expansion of nodes consumes more computer memory in order to store the information associated with each node, and more time, e.g., to calculate node expansions and points. Accordingly, efficient alternatives to blind-search methods are preferred. Heuristics may be applied to help focus the search, based upon special information that is available about the problem being represented by the graph. One way to focus the search is to reduce the number of successors of each expanded node.
  • Another way to focus the search is to modify the order in which the nodes are expanded so that the search can expand outwardly to nodes that appear to be most promising.
  • Search algorithms which modify the ordering of node-expansion are called ordered search algorithms.
  • Ordered search algorithms use an evaluation function to rank the nodes that are candidates for expansion to determine the node which is most likely to be on the best path to the goal node.
  • an f value is determined at each node n i available for expansion, where f is an estimate of the cost of a minimal cost path from the start node to the goal node constrained to go through node n i .
  • Each succeeding node having the smallest f value is then selected in sequence for expansion.
  • Fig. 20A illustrates a tree produced by an ordered-search algorithm applied to a blank workpiece that has four sections, which are to be bent upward to form four sides of a box, each side being represented in Fig. 20A by a corresponding number 1, 2, 3, and 4. Each numbered side of the box corresponds to a particular bend, including bend 1, bend 2, bend 3, and bend 4.
  • the blank workpiece corresponds to start node no which may also be called the root node associated with the initial state-description of the workpiece.
  • the successors of the start node n 0 may be calculated by expanding the start node (the root node) to form successor nodes n 1 , n 2 , n 3 , and n 4 .
  • nodes n 1 -n 4 correspond respectively to bend 1, bend 2, bend 3, and bend 4.
  • Node 1 is expanded to include successor nodes n 5 , n 6 , and n 7 which correspond respectively to bend2, bend3 and bend4, and an additional successor node n 8 which corresponds to a repositioning (i.e., a repo) of the robot gripper's hold on the workpiece.
  • Node 5 is expanded to include successor nodes n 9 and n 10 which correspond respectively to bend3 and bend4, and an additional successor node n 12 , which corresponds to a repo.
  • Node n9 is expanded to have successor nodes n 13 and n 14 , which correspond respectively to bend4 and a repo.
  • Node n 14 is expanded to have a successor node n15 which is the goal node, because it results in the final bend for the workpiece.
  • Bend sequence planner 72 preferably is configured to perform a best-first state-space search in order to develop a complete bend sequence to be performed by the bending workstation.
  • An ordered search algorithm utilizes an evaluation function to rank nodes that are candidates for expansion to determine the node which is most likely to be on the best path to the goal node, i.e., the node which is the best.
  • the first node corresponds to the flat part, e.g., as illustrated in Fig. 20A.
  • the best node which is on an OPEN list will be expanded, and the expanded node will be taken off OPEN.
  • all or a portion of the expanded nodes will be placed on OPEN.
  • the expanded nodes which are placed on OPEN will correspond to the remaining bend operations, minus those eliminated due to constraints.
  • twin nodes corresponding to each bend including a first twin node corresponding to operation of the bend while holding the workpiece from one side of the workpiece, and a second twin node corresponding to performing the same bend, but while holding the workpiece from the other side of the workpiece.
  • the expanded nodes which are placed on OPEN may also include one node that represents a repositioning of the robot gripper's grasp on the workpiece (i.e., a "repo").
  • certain levels of the search may be constrained so that they do not include a node for a repo.
  • bend sequence planner 72 will constrain the placement of a repo node on OPEN.
  • Figs. 20B and 20C each illustrate a simple example workpiece 16 having two faces 262, and one bend line 260.
  • each of Figs. 20B and 20C includes an accompanying diagram of a node expansion from the root node no to the first level of a search tree which includes two expanded nodes.
  • Fig. 20B shows two expanded nodes
  • Fig. 20C shows one expanded node and indicates that the other node has been constrained. Referring to Fig. 20B, since only one bend is to be performed on workpiece 16, only two nodes are shown.
  • the bend may be performed in accordance with node n 1 , whereby bend 1 is performed with face 2 being inserted into the die space of the bending workstation, or bend 1 may be performed in accordance with n 2 , whereby bend 1 is performed with face 1 being inserted into the die space.
  • the search tree illustrated in Fig. 20C only includes one node n 1 , and shows that while the parent node no might normally be expanded to include a second node, the second node has been constrained.
  • a node may be constrained by eliminating it from consideration as a possible operation within the bend sequence. Such elimination of a node may be accomplished by preventing an expansion from including the node, or by simply failing to place the node on the OPEN list.
  • Fig. 20D illustrates an example workpiece 16 having two co-linear bends, with bend lines 1 and 2.
  • the nodes that may be generated from this workpiece include the following: (1,2), (1,1), (2,2), (2,1), ((12),1), and ((12),2).
  • the holding faces are defined on each side of the first bend line of the co-linear bend.
  • Fig. 20E illustrates another example workpiece 16.
  • the holding sides for this co-linear bend (bending at lines 1 and 2 simultaneously) are defined in the following twin nodes: ((1 2) 1), ((1 2) 2).
  • the bend twin holding face is face 1, even though face 1 also extends to the other side of the bend line (i.e., even though it extends to a position which would be behind the die space during a bend). This is because of the convention noted above, which is used to choose the bend twin holding face.
  • Fig. 21 illustrates, in a simplified flow chart, an example embodiment of a state-search algorithm, comprising an ordered search algorithm, based on the algorithm disclosed by Nils J. Nilsson in "Problem-Solving Methods in Artificial Intelligence", which may be utilized by the bend sequence planner of the present invention in order to form a bend sequence to be utilized by a bending workstation.
  • a start node no is placed on a list called OPEN, and a function value f is set equal to 0.
  • step S12 a determination is made as to whether there is anything in the OPEN list. If the OPEN list is empty, the process is forwarded to step S14, and an error indication is given. If the OPEN list is not empty, as determined at step S12, the process will proceed to step S18.
  • step S18 the nodes placed within the OPEN list are checked, and the node having the smallest f value is removed from OPEN and placed on a CLOSED list. This node is called n i .
  • step S20 a determination is made as to whether the node n i is a goal node. If it is a goal node, the process is forwarded to step S22, where a solution path is generated by tracing back from node n i , through its pointer and the pointers of the previous nodes, to the start node n 0 . However, if node n i is not the goal node, as determined at step S20, the process will be forwarded to step S24.
  • each of the computed f values will be associated with their corresponding successor nodes n j that are not already on either the OPEN or CLOSED lists. Such successor nodes n j are then placed on the OPEN list, and pointers are directed from those successor nodes n j back to n i . For each successor node n j that was already on an OPEN or CLOSED list, an f value is associated with that successor node n j that is equal to the smaller of the f value just computed for that node and the f value already associated with that node.
  • step S24 The successor nodes n j on the CLOSED list who have their associated f values made smaller are placed on the OPEN list, and the pointers for those successor nodes n j are redirected to n i . After execution of step S24, the process will return to step S12.
  • Figs. 22A-22D illustrate a particular example embodiment of a bend sequence planning process to be performed by bend sequence planner 72 illustrated in Fig. 5A.
  • the bend sequence planning process is started upon receipt of a command to commence operation, e.g., as indicated in step S26, by proceeding on receipt of an FEL command to start planning. Once the process starts, and proceeds in step S28, one or more files corresponding to the parts to be produced are read by the bend sequence planner.
  • Such files may be in the form of a shape file including information such as geometric and topological information (a 3D data description of the part and a parallel 2D data description of the part corresponding to the 3D data description), labeled geometric features which are pertinent to determinations to be made by bend sequence planner, and a bend graph correlating bends to be performed with geometric and topological information.
  • geometric and topological information a 3D data description of the part and a parallel 2D data description of the part corresponding to the 3D data description
  • labeled geometric features which are pertinent to determinations to be made by bend sequence planner
  • a bend graph correlating bends to be performed with geometric and topological information.
  • step S30 the process proceeds to steps S30, S32, and S34, during which each expert is initialized. More particularly, the holding expert, the tooling expert and the motion expert are each initialized.
  • step S36 a list of bends is built, and calculations are performed regarding the various features of the parts. For example, a computation may be performed regarding what the lengths of bends are and which bends are co-linear.
  • step S38 an A* algorithm is initiated, including steps such as putting a root node no on an OPEN list, and setting an f value equal to 0. A determination is then made at step S40 as to whether the OPEN list is empty.
  • step S42 If the list is empty, the process will proceed to step S42, and exit with an error indication. Otherwise, if the OPEN list is not empty, the process will proceed to step S44, in which the node on the OPEN list with the smallest f value will be taken and placed on a CLOSED list.
  • the chosen node will be called n i for purposes of explaining the steps of the flow charts of Fig. 22A-Fig. 22D.
  • step S46 a determination is made as to whether node n i is a goal node. If node n i is a goal node, the process proceeds to step S48, where a solution path is generated. Otherwise, if n i is not a goal node, the process proceeds to step S50 which is shown at the top of Fig. 22C.
  • step S56 a finalize message is sent along with the bend sequence to each of the experts, and each of the experts is queried for final detailed information which is needed to complete the bend sequence plan.
  • step S58 the bend sequence planner will await a response from the tooling expert.
  • step S60 the setup of the bending workstation will be started.
  • step S62 the process will await a response from the motion expert and the holding expert. Once the complete motion expert and holding expert plans have been received, at step S64, the final plan will be forwarded to the sequencer of the system.
  • n i is not determined in step S46 to be the goal node
  • the process will continue at step S50 at the top of Fig. 22C.
  • node n i will be expanded to obtain its successor nodes n j .
  • the successor nodes will include bend twin nodes for each bend, i.e., two nodes corresponding to each bend, and an additional node for a repo, minus any nodes which are constrained from being successor nodes at the present level of the search.
  • step S52 a subplanning and cost assignment process is performed in step S52.
  • successors n j are each placed on the OPEN list, with the subplan information and cost information corresponding to each node being associated with each node in the OPEN list (e.g., by using pointers).
  • the process will then return to step S40 where a determination will be made as to whether the OPEN list is empty. If the OPEN list is empty, the process will exit with an error indication at step S42; otherwise the process will proceed to again execute steps S44, S46, S48, S50, S52 and S54.
  • Figs. 23A-23D illustrate the subplanning and cost assignment process which corresponds to step S52 in the bending sequence planning process illustrated in Figs. 22A-22D.
  • the subplanning and cost assignment process determines or formulates a subplan and incremental costs which correspond to each of the expanded/successor nodes n j which have not been eliminated as a viable node at the present level of the search due to constraints.
  • the process illustrated in Figs. 23A-23D will be performed.
  • a test will be performed for the permutability of node n j regarding the subplan and costs of the holding expert.
  • step S68 the process proceeds to step S70, where the subplan and costs are copied and associated with that successor node n j .
  • step S72 the bend sequence planner will query the holding expert for a proposed subplan, the incremental k cost, and the incremental h cost.
  • step S72 as soon as a cost of infinity has been evaluated by the holding expert, the present successor node n j will be aborted. Thus, the successor node n j will be discarded at the present level of the search, and the subplanning and cost assignment process will again start with the next available successor node n j .
  • step S76 at the top of Fig. 23B
  • step S76 at the top of Fig. 23B
  • step S76 another test for permutability will be performed regarding the tooling expert subplan and costs. If an equivalent node is found, as determined at step S78, the bend sequence planner will copy the subplan and costs corresponding to the equivalent node and associate the same with the present successor node n j . In the alternative, if an equivalent node is not found, the process will proceed to step S82 where the tooling expert will be queried for a proposed subplan, a k cost and an h cost. If a cost of infinity is evaluated, the present successor node will be aborted at step S84.
  • step S86 the bend sequence planner will await the results from the holding expert and the tooling expert.
  • the process will wait for the results of the holding expert and tooling expert queries, since such information is needed by the motion expert to do its subplanning and cost assignment computations.
  • step S88 a test will be performed for the permutability regarding the motion expert subplan and costs. That is, a test will be performed to determine if the subplan and costs that would be assigned by the motion expert are identical to those which have already been assigned to another node, the other node thereby being deemed an "equivalent" node to the present successor node n j being evaluated. If, at step S90, it is determined that an equivalent node has been found, the process will proceed to step S92, where the subplan and costs of the equivalent node will be copied and thereby associated with the present successor node n j .
  • step S94 the motion expert will be queried for a proposed subplan, a k cost and an h cost. If any of the costs are infinity, the present successor node will be aborted, proceeding to a next successor node and again commencing subplanning and cost assignment for the next successor node. Assuming that the proposed subplan and costs have been obtained, the process will proceed to step S98, where the results will be awaited from the motion expert. Additional processing may be performed to obtain a subplan and costs regarding different aspects of the system which will be related to performance of the overall bend sequence proposed by the bend sequence planner. In this regard, additional experts may be provided as indicated by the reference numeral S100. For example, Fig.
  • the subplanning and cost assignment process could be appropriately modified to include steps such as testing for permutability, querying the additional expert (e.g., sensing expert) for a proposed subplan and costs, and, at an appropriate location within the process, awaiting the results from the additional expert.
  • additional expert e.g., sensing expert
  • step S104 the f value will be adjusted based upon any heuristics which pertain to the successor node n j . In this regard, if it is a desired node, i.e., it has beneficial or desired heuristics which say that this node is preferable over other nodes, a value will be added to the f value. However, if the node is undesired, a value will be subtracted from the f value.
  • Fig. 24 illustrates an example flat workpiece 16, and several nodes expanded during the performance of a state-space search by the bend sequence planner illustrated herein. Various costs are shown which are assigned to the nodes throughout the search process.
  • flat workpiece 16 has two portions a, b which are to be bent to form flanges.
  • First flange a is placed in between two tabs c, d.
  • First flange a is to be bent along bend line 1
  • second flange b is to be bent along bend line 2.
  • the first node n 0 i.e., the root node, of the search tree corresponds to flat workpiece 16.
  • Successor nodes of node no include nodes n 1 and n 2 , which correspond, respectively, to bend lines 1 and 2.
  • the alternate node would likely be constrained in the present example. For example, it would likely not be possible to perform a bend along bend line 1 by inserting flange b into the bend press, since flange a is very short, and thus cannot be grasped by a robot gripper during execution of the bend.
  • the bend sequence planner may ask each of the holding expert, tooling expert, and motion expert for the incremental cost (i.e., h and k costs) corresponding to that node.
  • the costs that are assigned to node n 1 are illustrated in the box corresponding thereto as shown in Fig. 24.
  • a holding expert assigned a k cost i.e., the cost that it takes to move from the parent node no to the present node
  • the holding expert further assigned an h cost of 30.
  • the number 30 represents an approximate amount of time (30 seconds) which it will take to reposition the gripper's grasp on the workpiece 16 (i.e., to perform a repo). This value represents that the holding expert has predicted that one repo will be needed in order to complete the bend sequence associated with workpiece 16.
  • the h cost is a predicted cost to complete the bend sequence from the present node to the final goal node.
  • the costs assigned by the tooling expert include a k cost of 600 and an h cost of 600.
  • the k cost is the incremental amount of time (due to tooling) associated with performing the bend of that node.
  • a first stage must be placed on the die rail of the bending workstation.
  • An approximated time for installing the first stage is 600.
  • the incremental k cost (for tooling) from no to n 1 is 600 seconds.
  • the predicted additional cost from node n 1 to the goal node i.e., the h cost for tooling
  • the costs assigned by the motion expert include an incremental k cost of 5 (an estimated 5 seconds), equal to an approximated robot travel time in moving from n 0 to node n 1 .
  • the number of predicted repos is multiplied by 2, since two movements are required per repositioning.
  • the k value for the next node is calculated based upon the amount of time that it takes to move from the repo gripper to the appropriate stage for the next bend.
  • the alternate node at the first level of the search is node n 2 .
  • This node corresponds to bend line 2.
  • the incremental costs include k and h costs assigned by the holding expert, k and h costs assigned by the tooling expert, and k and h costs assigned by the motion expert.
  • the k and h costs assigned by the holding expert are 0 and 30 respectively.
  • the holding expert assigns a k cost of 0, because no repositioning is necessary to go from node n 0 to node n 2 .
  • a holding h cost of 30 is assigned because one repo is predicted to be necessary in order to complete all of the bends of the bend sequence, i.e., to get to the goal node. This becomes apparent when viewing workpiece 16.
  • the k cost assigned by the tooling expert again is 600, since the bend, being the first bend introduced in the search, will require at least one stage. 600 seconds is an approximated time for installing a stage, and thus is assigned as the incremental k cost to go from node n 0 to node n 2 .
  • the h cost assigned by the tooling expert is 600, since a predicted additional stage will be necessary to go from node n 2 to the goal node.
  • the motion expert assigns a k cost of 4, and an estimated h (future) motion cost of 12.
  • the k cost assigned by the motion expert for node n 2 is less than the k cost assigned for node n 1 . This is because bend line 2 is longer than bend line 1, and thus requires a larger stage.
  • a typical bending workstation such as the Amada BM100 workstation illustration in Fig. 1, it is preferred that longer stages be placed in the center of the die rail, and that shorter stages be placed off to the sides.
  • the calculated robot travel time without regard to the collisions, from the loader/unloader (L/UL) to the center stage in performing bend 2 is estimated to be 4 seconds, and less that it would take to get a stage positioned at the left side of the die rail which is where the smaller stage would be placed along the die rail. Since the h cost is calculated as a function of the present running average of the k cost calculated so far, the h cost is also a lower value of 12 seconds.
  • the respective total incremental costs performing bends 1 and 2, respectively, are 1250 and 1246. Accordingly, node n 1 has a total incremental cost of 1250, and node n 2 has a total incremental cost of 1246, the total cost being assigned by each of the experts queried by the bend sequence planner.
  • the only two nodes at the first level of the search included a node for performing bend 1, and a node for performing bend 2 (nodes n 1 and n 2 ).
  • the first level did not include a node for performing a repo. This is because the search is constrained so that the first bend to be performed at the first level after the root node does not include a repo. It would be unnecessary for a repo to be performed as a first step in the bend sequence, since the robot gripper can be placed anywhere at the start to correspond to any particular bend.
  • a repo is included as a possible node, in addition to the one or more bends which comprise the rest of the bends leading to the goal node.
  • the next level of the search includes nodes n 3 which corresponds to bend 1, and n 4 which corresponds to a repo before performing the next bend in the bend sequence.
  • the holding expert assigns a cost of infinity, since there are no available grasp regions that were used in performing bend 2 that can also be used to perform bend 1. If there was a grasp region that was used in order to perform bend 2 that could also be used to perform bend 1, then the robot gripper could be placed within that intersecting region, and the repositioning of the gripper would not be necessary when going from the completed bend 2 to bend 1 (i.e., from node n 2 to node n 3 ).
  • the holding expert has determined that there is no such intersection of grasp regions, and thus the incremental k cost for holding is infinity.
  • the predicted h cost is not even relevant, nor are any of the other costs which might be assigned by the other experts such as the tooling expert and the motion expert, since bend 1 cannot be performed at the present point in the bend sequence, without first performing a repo.
  • node n 3 is no longer considered, and the bend sequence planner proceeds to the repo node n 4 , and queries the respective experts for their assigned costs associated with that node.
  • the holding expert assigns a k cost of 30, which signifies that approximately 30 seconds will be needed to perform a repo at the present point in the bend sequence.
  • a predicted h cost of 0 is assigned by the holding expert, since it is predicted that no additional repos will be needed between the present node n 4 to the goal node.
  • the tooling expert After the holding expert assigns its cost, the tooling expert, upon being queried by the bend sequence planner, assigns a k cost of 600, which equals the approximate time (600 seconds) to install an additional stage which will be needed in order to perform bend 1 (along bend line 1), since the stage which was utilized to perform bend 2 (which has a length equal to the length of bend line 2) cannot be used to perform bend 1 since such a stage cannot fit between tab portions c and d of workpiece 16. No additional predicted stages or tooling change is expected by the tooling expert; and accordingly, the tooling expert assigns an h cost of 0 to be associated with node n 4 .
  • the tooling expert may initially determine a total initial h cost based upon the total amount of predicted stages that will be needed to perform the complete bend sequence, either at an initial point in the search before performing the search.
  • a total initial h cost is calculated to be 1200, since two predicted stages have been predicted to be necessary to perform bends 1 and 2 on workpiece 16.
  • the k cost is either 0 (with no extra stages needed) or 600 (if an additional stage is needed for the bend corresponding to the present node).
  • the cost assigned by the motion expert to correspond to node n 4 include a k cost of 8 and an h cost of 4.
  • the k cost is estimated to be twice the average preceding k cost, since two motions are needed in order to perform a repo.
  • One movement is needed to take the workpiece from a stage at which the workpiece was left from a previous bend to the repo gripper, and the second movement is to move the robot gripper to the repositioned location while the repo gripper is grasping workpiece 16.
  • the predicted h cost assigned by the motion expert for a repo node is the predicted additional costs needed to perform all future movements in the bend sequence.
  • h is estimated to be the h value calculated for a previous node n 2 minus the present k cost, and thus is estimated to be 4 seconds for node n 4 .
  • the total incremental costs are then added to the total of all previous k costs preceding that node (in this case repo node n 4 ).
  • all the incremental associated with node n 4 are added to a total previous k costs of 604 which were previously calculated in association with node n 2 , to obtain a total cost value of 1246.
  • the bend sequence planner will, in performing its state-space search, thus choose n 4 as the best node and will proceed with expanding that node to form its successor nodes.
  • the successor nodes of repo node n 4 include node n 5 .
  • node n 5 is the goal node, since it results in the workpiece 16 having all of its bends completed to form a 3D part.
  • the costs determined by the relative experts include a presumed holding k cost of 0, a calculated tooling k cost of 600, and a calculated motion k cost of 4. Since the present node n 5 is known to be the goal node, no h costs are calculated. The previous total k costs 642 seconds.
  • 642 is added to the k cost for tooling of 600 and the k cost for motion of 4 to be equal a total f value of 1246.
  • Such an f value is the cheapest f value among the nodes still left on OPEN. Accordingly, this node will be checked to see if it is a goal node, and if it is a goal node, the solution path will be generated to include (in order) bend 2 which corresponds to node n 2 , a repo which corresponds to node n 4 , and bend 1 which corresponds to node n 5 .
  • a test is performed for the permutability of that node regarding the subplan and costs for each expert. For example, in step S66 shown at the top of Fig. 23A, a test is performed for the permutability of a particular successor node n j to determine if it is merely a permutation of another node, and thus has an equivalent set of subplan and costs. If this is the case, it would be wasteful to again ask the holding expert for a proposed subplan and associated k and h costs, since these parameters are already known, and can be obtained by merely referring to the other equivalent nodes.
  • 26 illustrates a graph of compared histories of nodes b6' and b6, which have been generated by the bend sequence planner in performing its state-space search.
  • steps S66 Fig. 23A
  • S76 Fig. 23B
  • S88 Fig. 23C
  • a test will be performed for the permutability of that node with any other nodes in the search tree regarding the holding expert's subplan and costs, the tooling expert's subplan and costs and the motion expert's subplan and costs, respectively.
  • a node such as node b6 illustrated in Fig. 26 will be compared to another node in the search tree, such as node b6', also illustrated in Fig. 26.
  • the history of node b6, which includes nodes b2, r1, b4, b3, r2, and b5 is compared to the history of b6', which includes b2', r1', b3', b4', r2' and b5'.
  • the method to be used to determine whether one node is a permutation of another will vary.
  • an analysis can be performed of the various permutations of nodes, and the various subplans and costs that can be associated with each node at various levels of the search, in order to determine under what conditions a node is a mere permutation of another node in the search.
  • an appropriate method may be formed for determining whether a node is a permutation of another node, in terms of the subplan and costs assigned for the node.
  • Figs. 29-31 respectively include charts which depict the dialogue between the bend sequence planner and the holding expert, tooling expert, and motion expert of the illustrated embodiment planning system 71 as shown in Fig. 5A.
  • Fig. 29 which illustrates the dialogue between bend sequence planner 72 and holding expert 82
  • query arrows Q1, Q2, Q3, Q4 and Q5 are illustrated to represent a query message being forwarded from the bend sequence planner 72 to holding expert 82.
  • response arrows R1, R2, R3, R4, and R5 are illustrated to represent response messages from holding expert 82 to bend sequence planner 72. While the queries and responses are indicated in Fig.
  • bend sequence planner 72 forwards an initial query Q1 to holding expert 82, which includes, among other things, a start command, and a file name for the part to be produced.
  • This query Q1 would be forwarded utilizing a VERB "plan " (which is utilized to initialize a module for planning a part).
  • the holding expert Upon receipt of query Q1, the holding expert then performs an input operation indicated by Il, which includes reading an appropriate file which includes geometric, topological, feature information, and other information regarding the parts to be produced. After the part is read, initial planning steps will be performed, as indicated in block P1.
  • holding expert 82 will perform gripper selection, which includes picking a robot gripper, and which includes picking a temporary repo gripper. In addition, holding expert 82 will predict the minimum number of repos that will be needed to complete the overall bend sequence. After performance of the initial planning steps in P1, holding expert 82 then sends the resulting information back to bend sequence planner 72 via a response R1.
  • the response includes a savelist which includes a list of names of attributes to be saved by bend sequence planner 72.
  • the savelist further includes, along with each attribute name, the parameters and values accompanying each attribute to be saved by bend sequence planner 72.
  • the attributes to be saved by bend sequence planner 72 at this point include the selected robot gripper, the temporarily selected repo gripper, and the values indicative of the minimum predicted number of repos which will be necessary to complete all of the bends of the bend sequence.
  • a query Q2 is sent to holding expert 82 (e.g., at step S72 of the bend sequence planning process illustrated in Fig. 23A).
  • the query Q2 includes bend sequence information, and a request for a proposed subplan, a k cost and an h cost associated with that particular node.
  • a "get" FEL command may be utilized to perform this query.
  • holding expert 82 will then perform planning steps indicated in block P2, which include predicting the number of repos which will be needed after performance of the presently proposed bend-related operation, determining the grasp location (i.e., the location at which the robot should grasp a workpiece in order to perform the presently proposed bend), and potential repo locations (for the repo gripper's grasp on the workpiece), and will also determine k and h costs associated with the particular proposed bend-related operation (which would include either a bend or a repo).
  • the grasp location i.e., the location at which the robot should grasp a workpiece in order to perform the presently proposed bend
  • potential repo locations for the repo gripper's grasp on the workpiece
  • k will either be equal to 0 or infinity, 0 indicating that no repo is necessary at the present node, and infinity indicating that there are no available places for the robot to grasp the workpiece without first performing a repo.
  • the h value will be equal to 30 (an estimated amount of time it takes to perform a repo) times the predicted number of repos from the present node to the goal nodes.
  • k will be equal to 30, if the repo is possible, or infinity if a repo cannot be performed at the present level of the search for the present node.
  • the h cost will be 30 times the predicted number of repos which will need to be performed after performance of the present node bend-related operation.
  • bend sequence planner 72 After performance of processing in relation to query Q2 and response R2, bend sequence planner 72 will then query various other experts including tooling expert 80 and motion expert 84, in order to obtain their respective subplans and costs, and repeatedly will query each of the experts in association with each node generated during the search in order to form a complete bend sequence plan which includes nodes from the start node to the goal node.
  • bend sequence planner 72 will forward another query Q3 to holding expert 82 which includes a request for the suction cup plan, again utilizing the "get" verb of FEL.
  • holding expert 82 will perform suction cup planning as indicated by block P3.
  • Suction cup planning will include a determination of what locations along the workpiece loader/unloader may place its suction cups during loading and unloading of the workstation.
  • holding expert 82 will respond with response R3 to bend sequence planner 72.
  • Bend sequence planner 72 will subsequently again query, by query Q4, holding expert 82, for the final repo gripper that will be used and the location of the repo gripper on the workpiece for various stages of the bend sequence.
  • the "get" verb of FEL may be used for this query.
  • holding expert 82 will perform the planning indicated in block P4, which includes repo planning to be performed after the search.
  • holding expert 82 chooses a true repo gripper to be utilized in execution of the resulting bend sequence plan, and finalizes the repo position based upon the chosen repo gripper.
  • holding expert 82 will forward a response R4 to bend sequence planner 72.
  • bend sequence planner 72 will further query holding expert 82 for a backgage plan.
  • holding expert 82 will perform backgage planning as indicated by block P5, and will respond to bend sequence planner 72 with the appropriate backgage plan in response R5.
  • bend sequence planner 72 will have queried the motion expert 84 for its final plan information, and will await, before execution of the plan, the results of the final motion plan from motion expert 84. After receipt of the final motion plan from motion expert 84, bend sequence planner 82 will then proceed to forward the final plan to sequencer 76.
  • bend sequence planner 72 In the illustrated dialogue between bend sequence planner 72 and tooling expert 80 in Fig. 30, several queries are illustrated from bend sequencer planner 72, indicated by query lines Q11, Q12, and Q13, and several responses are illustrated by response line R11, R12, and R13.
  • first query line Q11 As indicated by the first query line Q11, at some point in time before commencing its search (e.g., at step S32 in the bend sequence planning process illustrated in Fig. 22A), bend sequence planner 72 will command tooling expert 80 to start its processing, and will forward the name of the part to be produced with the use of a "plan" verb in FEL.
  • tooling expert 80 Upon receipt of query Q1, as indicated by input line I2, tooling expert 80 will then read an appropriate part file.
  • tooling expert 80 will perform various planning steps as indicated by blocks P11, P12 and P13. These planning steps include selection of a tool profile, bin-packing, and performing a calculation of an initial h value (which corresponds to the total number of predicted stages that will be needed to perform all of the bends of the bend sequence).
  • the bin-packing algorithm comprises the selection of tool segments that will together add up to the appropriate stage length for each stage to be utilized by the bending workstation in performing the bends of the bend sequence.
  • the bend sequence planner 72 will commence it search. Once the search is commenced, and after the information has been gathered from holding expert 82, bend sequence planner 72 forwards a query Q12 to tooling expert 80, which includes the bend sequence at that point of the search and a query for the subplan and associated k and h costs. The verb "get" in FEL is utilized for this query. Tooling expert 80 then performs planning steps, as indicated by planning block P14, which include picking of a stage length to correspond to a bend and a location along that stage where the bend should be performed, arranging the stages, calculating the k and h costs, and performing fine motion planning.
  • tooling expert 80 responds to bend sequence planner 72 via response R12, and forwards the k and h costs and the associated subplan information to bend sequence planner 72.
  • a savelist is also included in response R12 which indicates information and attributes that should be saved by the planner.
  • Subsequent queries and responses may be exchanged throughout the search, with tooling expert 80 and with other experts 82 and 84 before the search is finished.
  • a query Q13 instructing the tooling expert to finalize will be forwarded to tooling expert 80.
  • Tooling expert 80 will then perform its appropriate final processing, and return, via response R13, any final information to bend sequence planner 72.
  • bend sequence planner 72 requests final information and final processing to be performed by motion expert 84 and will await the results thereof. Once the final motion planning results have been obtained by motion expert 84, bend sequence planner 72 will compile all information to form a final plan, and will forward the same to sequencer 76.
  • bend sequence planner 72 communicates with motion expert 84 before, during and after performing a search, in the form of queries and responses which may include the queries indicated by query lines Q21, Q22 and Q23, and respective response lines R21, R22 and R23.
  • queries and responses which may include the queries indicated by query lines Q21, Q22 and Q23, and respective response lines R21, R22 and R23.
  • a first query Q21 may be forwarded to motion expert 84 which includes a start command, and the name of the part to be produced.
  • motion expert 84 Upon receipt of query Q21, motion expert 84 will then input the appropriate part file and a channel file which represents all of the free space channels through which the part and the robot may be manipulated in performance of the various bends and operations of the bend sequence. This input is indicated by 13.
  • motion expert 84 will send a response R21 to bend sequence planner 72, indicating, essentially, that the information was read in and acknowledging that it is ready for the next query by bend sequence planner 72.
  • the state-space search of the bend sequence planner 72 will commence.
  • bend sequence planner 72 will query holding expert 82 for various information while performing the first level of the search, then query tooling expert 80, and thereafter send a query Q22 to motion expert 84.
  • Query Q22 includes information about the bend sequence, the gripper location and the bend locations on the stages (in the form of a bend map). This query may be sent to motion expert 84 by using a "get" verb in FEL.
  • motion expert 84 Upon receipt of query Q22, motion expert 84 will perform processing in processing block P21, and thus will develop a subplan and determine the k and h costs for performing the bend proposed by bend sequence planner 72 at that particular point in the bend sequence. The resulting k and h costs and subplan are returned to bend sequence planner in response R22. Afterward, additional processing by other experts 80, 82, and by motion expert 84 may be performed in order to complete the search.
  • bend sequence planner 72 will send an additional query Q23, which includes a finalize command.
  • bend sequence planner 72 will forward information to motion expert 84 so that motion expert 84 may perform all final planning operations.
  • Such forwarded information would include the bend sequence, the gripper locations for each bend in the sequence, the repo locations for each repo to be performed, the bend maps corresponding to the bends of the bend sequence, and all fine motion plans which have been developed by tooling expert 80, in order to bring the workpiece into and out of the die space when performing each bend in the bend sequence.
  • Motion expert 84 utilizes that information to perform the processing indicated in processing block P22. More particularly, motion expert 84 will figure out the various starting and finishing points in order to develop a gross motion plan.
  • a search algorithm is then performed in order to form paths between the gross motion starting and finishing locations. Then, the resulting gross motion paths are linked with the fine motion paths so that a complete motion scheme is formed, commencing with acquiring the workpiece from the loader/unloader during loading of the workstation, bringing the workpiece to each of its bends, and finally bringing the finished workpiece to the loader/unloader to be unloaded from the workstation.
  • bend sequence planner 72 may compile the complete plan, and forward the same to sequencer 76 for execution.
  • Fig. 32 illustrates a flow chart of an example process for performing robot gripper selection. This process is performed, e.g., in planning block P1 in Fig. 29.
  • a library of grippers is read in.
  • the process prunes obviously bad grippers, e.g., if they have certain dimensions which are inappropriate for the type of work being performed by the bending workstation.
  • a minimum number of repos for each gripper is predicted.
  • the one or more grippers having the smallest predicted number of repos is selected.
  • step S136 among the selected grippers, all of the grippers having the largest width are selected.
  • the grippers with the shortest knuckle height are selected. If there is only one gripper having the largest width among the selected grippers, then that gripper will be selected and no further determination is needed as to the length of the gripper or as to the knuckle height of the gripper. Similarly, if several grippers have the largest width among the select grippers, but only one gripper has the smallest length, then that gripper will be selected and no further determination will be needed as to the knuckle height of the gripper. If there are several grippers left that have an equal shortest knuckle height, as determined in step S136, then any one of those grippers may be chosen. Thereafter, in step S138, the chosen gripper is returned to the holding expert.
  • Such a prediction of the minimum number of repos, before the search, can be performed by utilizing the exemplary process illustrated in Fig. 34A.
  • the goal of the process depicted in Fig. 34A is to, for a given robot gripper and a given part, predict the minimum number of repos that will be needed in order to form the complete 3D part.
  • information is needed regarding both the 2D part, and 3D part (the completely formed bent part).
  • discrete points are generated around a periphery of a part of a 2D representation of the part.
  • Fig. 33A Such discrete points, located a set distance from the edge of the part, are illustrated in Fig. 33A.
  • the granularity shown in Fig. 33A is merely for the purpose of explanation of the algorithm, and does not necessarily reflect a preferred granularity.
  • the granularity of the discrete points may be varied in order to obtain an optimum accuracy, while not sacrificing the speed of the search process.
  • a bend set including all of the possible bends for that robot grasp position will be identified in step S142, assuming that the part is still flat, (in 2D) and that the part is at the L/UL. This is repeated for each discrete point around the periphery of the 2D part 16a (e.g., as shown in Fig. 33A), and all bend sets for each corresponding robot grasp point are identified.
  • step S144 a determination is made as to the minimum number of unions of the bend sets determined in step S142 that will form a complete set of bends (i.e., all of the bends of the bend sequence). This minimum number of unions will be identified as a 2D minimum number of repos R2.
  • step S146 the discrete points are generated around the periphery of a 3D part 16b (e.g., as shown in Fig. 33B). It is noted that the granularity shown in Fig. 33B is only shown by way of example, and does not necessarily represent the preferred granularity for performing the present algorithm.
  • the appropriate granularity for the generation of points around the outer periphery of the part may be modified in accordance with the desired accuracy and speed of the algorithm.
  • the corresponding bend set i.e., all of the possible bends that may be performed when the robot is grasping the part at that location
  • the minimum number of unions required to get a complete set of bends i.e., all of the bends of the bend sequence
  • R3 represents the minimum number of 3D repos.
  • step S148 all of the possible sets of bends in grasping at the respective discretized X positions on 3D part 16b are formed assuming a particular gripper, and further assuming that the 3D part is located at the repo station.
  • the values R2 and R3 are returned to the algorithm for selecting the robot gripper (e.g., as disclosed in Fig. 32) and to the holding expert.
  • the value R3 represents an upper bound number of predicted repos, since it is more difficult to hold the workpiece when it is completely bent, i.e., a 3D part, than it is to hold the workpiece in performing bends when it is a flat part.
  • the value R2 represents a lower bound number of predicted repos.
  • the selection of robot gripper algorithm and the holding expert may each utilize either the lower value R2, the upper value R3, or a combination of the two in performing their calculations and determinations.
  • the lower number R2 may be first considered. If there are more than one grippers with an equal smallest predicted number of repos R2, but with different values R3, then the grippers with the smallest value R3 may be selected. These selected grippers, if more than one, would then be further evaluated in accordance with step S136, as shown in the flow chart of Fig. 32.
  • Fig. 35A illustrates a process for predicting the minimum number of repos which can be used during the search.
  • the algorithm for predicting the minimum number of repos used before the search did not include an evaluation of intermediate parts, in order to save time.
  • the algorithm depicted in Fig. 35A also considers a formed intermediate part, and the variations of the part as it moves throughout the various bends.
  • a first step S154 an intermediate part is formed, by calling an appropriate function in a geometric modeling library.
  • the intermediate part includes all of the bends in the bend sequence so far up to the present node of the search.
  • step S156 discrete points are generated around the periphery of the intermediate part, in a manner similar to that described in the process of Figs. 34A-34B, and in a manner similar to that illustrated in Figs. 33A and 33B.
  • step S158 a bend set is determined for each grasp location point. In other words, a determination is made as to all of the possible bends that may be performed while the robot gripper is grasping the part at each discretized point.
  • step S160 a determination is made as to the minimum number of unions of the bend sets generated in step S158 needed to form a complete set of bends (i.e., all of the bends of the bend sequence). This number is called Ri.
  • Ri the minimum number of unions of the bend sets generated in step S158 needed to form a complete set of bends (i.e., all of the bends of the bend sequence).
  • step S162 discrete points are generated around the periphery of the 3D part.
  • a bend set i.e., the possible bends that may be performed for each gripper position along the discretized points
  • the minimum number of unions of the bend sets is then determined which would be necessary to form a complete set of bends (i.e., all of the bends of the bend sequence). That minimum number of unions is referred to as R3.
  • a low h cost Ri(c) and a high h cost R3(c) are assigned and returned to the planner.
  • the cost values Ri(c) and R3(c) are estimates of the amount of time it takes to perform a repo times the minimum number of repos (Ri and R3, respectively).
  • the process for predicting the minimum number of repos during the search may send the values Ri and R3 themselves.
  • Fig. 36A illustrates an example process for determining the robot grasp locations, as performed in planning block P2 in the chart depicted in Fig. 29 by holding expert 82.
  • a first step, S170 an intermediate part (having the bends corresponding to the present node of the state-space search of the bend sequence planner) is constructed.
  • step S172 all edges which are not appropriate for grasping are rejected. For example, an edge may be rejected if it is not a face which is parallel to the robot's XY plane. In addition, an edge may be rejected if it is inaccessible by the robot gripper when the part is loaded in the die space.
  • the edge may be rejected if the edge is too close to the die, so that the robot would collide with the tooling before and/or during the bend operation.
  • the edge may also be rejected if grasping the workpiece on such an edge would cause the robot to be outside of its work space.
  • step S174 for each non-rejected edge, every vertex is transformed from sheet coordinates to edge coordinates.
  • Fig. 37 an illustration is provided in order to define an example set of sheet coordinates X s and Y s on a workpiece 16 having bend lines 1, 2, 3, and 4, which may be transformed to edge coordinates X e and Y e which correspond to the edge of workpiece 16 which is next to bend line 1.
  • each edge is discretized ties into points along the X axis in step S176.
  • grasp lines are generated which extend along the Y axis.
  • several process steps are performed. For example, referring to Fig. 38, for a discretized point x p , a (broken) grasp line 306 is formed along the Y axis.
  • an initial Y value Ys is proposed which is set at a distance from the edge (e.g., 3 mm).
  • the gripper is oriented to be normal to the X axis in edge coordinates.
  • a determination is then made as to whether or not the point Ys is out of the robot's work space, while the workpiece is at the loader, the repo station, or at one of the stages. If this is the case, a new point along a line corresponding to the discretized ties point Xp and normal to the edge is found that is within the work space. For the first valid Yp, a determination is made as to whether Yp is beyond the gripper's maximum reach. If so, the value Yp is rejected. In addition, a determination is made as to whether or not the gripper can make good pad contact with the part if the gripper is at the position Yp.
  • Yp is rejected. New values for Yp are proposed, until line 306 reaches a first maximum location at which the robot can grasp the part, that first maximum position being Yf. This distance is defined by the fact that pads cannot have good contact any more due to holes or due to a boundary in the part. For example, a maximum position Yf is found right before a first hole 307 in the workpiece 16 shown in Fig. 38. The next viable or potential Yp is then found along the line running perpendicular to the edge and is defined as a new initial or starting position Ys'.
  • Y values Yp are then proposed and tested until an additional final position Yf' is found due to limits because the pads cannot have good contact or due to the fact that the part has a boundary at that location.
  • Yf' is determined to be just before second hole 308. This process is repeated until the end of the line 306 reaches the gripper's maximum reach or the boundary on the opposite side on workpiece 16. Thus, an additional line segment extending from Ys" to Yf" is generated.
  • a common grasp area is defined for the present bend in the search, and is defined to be the intersection of the current grasp lines with the grasp lines determined for previous bends since the last chosen repo in the search.
  • a k cost of 0 is assigned if the intersection is not equal to 0, and a k cost of infinity is assigned if the intersection is 0. This signifies that the present bend cannot be performed since the grasp areas needed to perform the bend are not common with the previous bend.
  • a temporary grasp location is selected within a defined common area.
  • final grasp locations are selected for the bends preceding the repo, since it is known that the grasp location will not change any further for that set of bends.
  • a final grasp location is selected such that a large repo are is generated.
  • Fig. 39 illustrates the evolution of the common grasp area as determined throughout a search, as calculated by a determined robotic grasp locations process, e.g., as illustrated in Figs. 36A-36B.
  • the grasp area for bend 1 is first determined as illustrated in view A. Then, with bend 1 having been already performed, and the corresponding flange being bent (indicated by the cross-hatched lines in view B), the potential grasp regions which can be utilized to perform bend 2 are determined as illustrated in view B. The intersection of the regions in views A and B is then determined as illustrated in view C. Then, bend 2 is performed (indicated by cross-hatched lines in view D), and the total available grasp regions which may be utilized to perform bend 3 are determined as shown in view D.
  • Each bend, which has already been performed, is indicated by cross-hatched lines being placed on the flange that is bent.
  • the grasp regions are indicated by a solid black line.
  • Fig. 40 illustrates first and second views of a workpiece 16, the views showing the grasp line regions determined before performing a first bend, and before performing a second bend, respectively.
  • the grasp line region 309 comprises a certain large area of the workpiece 16.
  • the lower view illustrates the intersection of the grasp line region (i.e., the grasp area) shown in the top view which can be utilized to perform the first bend and a grasp line region (not shown) which may be utilized to perform the second bend.
  • grasp line region 310 is a small subset of the grasp line region 309, and may be utilized as a grasp location to perform both the first bend and the second bend.
  • Fig. 41A illustrates an example embodiment of a process for determining the repo gripper locations which will be performed during repo planning after the search as indicated by planning block P4 in Fig. 29.
  • a first step S184 an intermediate part is constructed.
  • the edges which are not appropriate are then rejected in step S186.
  • the process may reject an edge if it does not correspond to a face which is parallel to the robot's X-Y plane.
  • the steps following step S186 are performed.
  • the intermediate part is transformed from sheet coordinates to edge coordinates.
  • the edge of concern is discretized along the X axis (in a manner similar to that illustrated in Figs.
  • grasp lines are generated along the Y axis, by generating points along the Y axis from a first point Ys (e.g., 3 mm) to the gripper's maximum reach along the line which is placed on the discrete X point. For every point along that line, if the repo gripper interferes with a previous robot gripper location, that Y location is rejected. In addition, for each Y position, if the repo gripper interferes with any portion of the part, that Y position is rejected. In addition, if there is no good pad contact between the repo gripper and the part, that Y position is rejected. A line is thus drawn as shown in Fig.
  • a final position Yf which is a first maximum position that the repo gripper may grasp the part until it hits a boundary portion (e.g., a hole in the part). Additional sets of initial and final positions Ys and Yf are formed until the repo gripper reaches its maximum reach (e.g., at Yf" as shown in Fig. 38), in a manner similar to that performed in the process for determining the robot's grasp locations as disclosed in conjunction with Figs. 36A and 36B.
  • a final repo location is assigned (in consideration of previous and current robot gripper location) when the search reaches the goal or another repo becomes necessary.
  • Fig. 42 illustrates an example embodiment of the process for performing repo gripper selection before the search. This may not be actually implemented.
  • a library of grippers is read, and in a second step S200, a conservative repo gripper is selected.
  • a conservative repo gripper is defined as a gripper which is narrow and short, and is capable of holding the part (in either 3D or 2D shapes).
  • the selected repo gripper is a temporary solution, since a final repo gripper selection will be performed after the search is completed.
  • the repo gripper selection after the search is illustrated in Figs. 43A-43B.
  • a first step S202 all the intermediate part geometries for the various bends throughout the bend sequence are constructed.
  • step S204 grippers are pruned, which are deemed inappropriate due to obvious reasons (e.g., they cannot grasp a part because of insufficient dimensions).
  • step S206 available repo grippers are identified based upon two robot grasp locations which include an initial robot grasp location before the repo and a repositioned robot grasp location. Each of these positions has been already determined in the search process. If the previously determined temporary repo position, determined during the search, could be improved upon in view of the repo grippers that are identified as available, then the position is adjusted.
  • step S208 if there are more than one available repo grippers (after pruning), then the repos with the largest width are selected. If there are more than one repo grippers with the largest width, then the ones with the smallest length are chosen. If there is more than one repo gripper with the smallest length, then the one with the shortest knuckle height is chosen. If there are several repo grippers with the same smallest knuckle height, then any one of those is chosen. Currently, a repr gripper is selected such that it allows a larger robot gripper to be selected and it guarentees a successful generation of repo gripper locations. The width of a repo gripper is determined by the dimension of possible area of 3D part for grasping. The knuckle height of a repo gripper is determined to be taller than the minimum flange height of 3D part.
  • a bin-packing algorithm is performed before the search is started.
  • a plan is produced that specifies how the segments should be put together to form each stage in a list of stages to be chosen from.
  • Fig. 44 illustrates an example bin-packing algorithm.
  • the process builds a list of bend line lengths, and forms a stage length list having stage lengths equal to the lengths of the bend lines to be formed on the workpiece.
  • the process builds or reads a list of available segment lengths which may be chosen from in order to form the stages in the stage length list.
  • step S212 an A* search is performed in order to determine a combination of segments which could be used to form the particular stage.
  • step S214 the process returns a solution set of tool/die segments to the tooling experts.
  • the initial node no is expanded to include a plurality of nodes at the first level of the search, each of the expanded/successor nodes at the first level corresponding to one of the available segment lengths (i.e., lengths of a tool punch and corresponding die segments) For example, if the available tool segment lengths are 10 mm, 15 mm, 22 mm, 40 mm, 80 mm and 160 mm, the nodes at the first level would correspond to each of those segment lengths.
  • the available segment lengths i.e., lengths of a tool punch and corresponding die segments
  • the k cost assigned for each successor node is the length of the segment corresponding to the present node and the h cost is set equal to the length of a remaining portion of the stage which is yet to be formed by the segments (i.e., how far the search process is from the goal).
  • Figs. 45-46 illustrate how the h cost that is assigned by the tooling expert throughout execution of the search, and forwarded to the bend sequence planner 72 (in response R12, as shown in Fig. 30), is calculated.
  • the tooling h cost is determined as a function of the total number of predicted stages that will be needed to perform all of the bends in the bend sequence.
  • h initial is an initial h cost equal to the total number of predicted stages needed to perform all bends of the bend sequence multiplied by an approximate amount of time (e.g., 600 seconds) needed to install each stage
  • k' TE for n j is the summed tooling k costs from node no to node n j .
  • a bend "test strip" 370 is laid across each bend line of the 2D representation of the workpiece. In each of the examples shown in Fig. 45, such a bend "test strip” 370 is laid across the bend line which is darkened.
  • Fig. 46 illustrates an example flow chart of the steps that may be performed in order to determine the initial tooling h-cost (h initial ), which is the total number of predicted stages needed to perform all of the bends on the workpiece.
  • h initial the initial tooling h-cost
  • a first step S216 a first stage length which is equal to the length of the longest bend line is placed within the set of assigned stages.
  • a test is performed for each bend line, by performing step S218 and the steps following step S218 for each bend.
  • step S220 If a difference value, equal to the total number of faces after placing the test strip over the bend line minus the total number of faces before the test strip, is less than or equal to 3, then no extra stage is needed. Otherwise, an extra stage is needed.
  • step S220 if an extra stage is needed (i.e., predicted), the longest stage (from the stage list) that can be used to perform the bend being tested is assigned, i.e., placed in the set of assigned stages. Then, a determination is made in step S222 as to whether the newly assigned stage is equal to a stage already in the set of assigned stages. If the newly assigned stage is already in the set of newly assigned stages, the newly assigned stage is not appended to the set, as indicated in step S226.
  • step S224 the process returns from either of steps S224 and S226 to step S218, if there are additional bend lines which need to be evaluated. Once all the bend lines have been evaluated by the process, the process proceeds to step S228, where the initial tooling h_cost is set to the product of 600 and the predicted number of stages (which is the total number of stages which have been placed in the set of assigned stages).
  • Fig. 47A illustrates a tool selection process overview which forms part of the tool profile selection planning block P11 in Fig. 30.
  • the process begins at the bend sequence planner in step S471, and proceeds to the tooling expert (tooling module) which operates in step S472.
  • the tooling expert forwards the part's geometric model, bend-graph data, and a tool library to a tool filter module.
  • the tool filter module determines a selected die, die-holder, die-rail and a list of feasible punches. In determining such information, the tool filter module performs several steps for each bend to be performed on the workpiece as indicated by the bend-graph data.
  • the tool filter module reads necessary data for the bend, and selects the die, die-holder, and die-rail based upon tonnage, V-width, angle and inside-radius requirements. The tool filter module then prunes the list of punches (to form a list of feasible punches) based upon tonnage, tip radius and tip angle constraints.
  • step S473 The process then returns to the tooling module in step S473, which then forwards the part's geometric model, bend-graph data and a list of feasible punches to a profile select module.
  • step S474 the profile select module selects the punch and punch holder to be utilized by the bending apparatus. In performing the profile selection, for each bend, the profile select module selects only those punches from the feasible list whose profile matches the geometry of the part. Punches with matching profiles will not collide with the part during the bending process. The profile select module then selects the best punch and punch-holder accordingly. The appropriate selected punch and punch holder are then returned to the tooling module which continues its functions at step 5475.
  • the tool filter module reads the following data: the desired inside radius (IR) of each bend; the part material thickness (T), the part material tensile strength; the minimum adjacent flange length (the minimum/preferred minimum length (height) of the shorter flange which runs along the bend line of the particular bend of concern); the bend length and bend angle; and a tool library (the tool library includes inverted profiles of the punches which can be used).
  • the tool filter module performs the following steps:
  • the tonnage-meter requirement for the v-width and T may be computed using a force chart and tonnage equations provided by Amada in their press brake tooling catalogues.
  • the tonnage-per-meter value may be calculated using the bend chart and tonnage equations provided in the text entitled “New Know-how on Sheet-Metal Fabrication Bending Technique,” written by the Amada Sheet Metal Working Research Association, Machinists Publishing Company, Ltd., First Edition (May 15, 1981)
  • the final (finished) 3D model of the part is aligned in relation to the appropriate tooling in a position in which it would be in the bending press after completion of the bend being evaluated. Then, for each bend:
  • tool filter module and profile selection module calculations may be performed either before, during or after the search is performed by the bend sequence planner.
  • Figs. 47B-47C illustrate a stage planning process which picks a stage and a location along the stage at which the workpiece will be loaded when performing a particular bend in the bend sequence, such planning being indicated in block P14 of the dialogue diagram shown in Fig. 30.
  • a first step S230 an intermediate part model of the part is formed (with the part having the bends up to the present bend in the bend sequence).
  • step S232 the biggest non-evaluated stage is chosen from the stage list (of available stages).
  • step S234 the present bend in the search is simulated with tooling expert (TE) collision checking, with the part being loaded at onto the tooling stage at a center position with respect to the stage.
  • step S236 a determination is made as to whether or not there was a collision during simulation of the bend. If there was a collision, the process proceeds to step S238, where the bend being evaluated in the search is simulated with TE collision checking while the part is loaded at the left side of the tooling stage, with the left end of the bend line being placed just to the left of the left side of the tooling stage. If a collision is then determined in step S242, the process proceeds to step S246.
  • TE tooling expert
  • step S236 If, however, a collision is not determined to have occurred in step S236, the position at which the workpiece will be loaded onto the stage is set to the center position in step S240, and the process proceeds (via connector B) to step S254 which is shown in Fig. 47C.
  • step S242 If a collision is not determined in step S242, after simulating the bend with the part positioned on the left side of the stage, the process proceeds from step S242 to S244, where the position for loading the workpiece on the stage is set to the left position. Then the process proceeds directly to step S254 (via connector B).
  • step S246 the bend is simulated with TE collision checking with the part positioned at the right side of the tooling stage (e.g., as shown in Fig. 48B), with the part being placed on a tooling stage so that the right end of the bend line is placed just to the right of the tooling stage while the bend is performed. If a collision is determined to have occurred during this simulation, the process proceeds to step S252. If no collision occurred during this simulation as determined in step S248, the process proceeds from step S248 to step S250, wherein the loading position is set to the right position, before the process proceeds to step S254.
  • step S248 If a collision did occur as determined at step S248, the process proceeds to step S252, wherein the chosen stage (chosen in step S242) is disregarded, and the process proceeds (via connector C) to step S232 at the top of Fig. 47B. It is noted that the next non-evaluated biggest stage from the stage list will be chosen in step S232 at this point. However, the stage planning process may be designed so that it will go from a "failed" biggest stage straight to a stage having a length approximately equal to the bend line of the particular bend being evaluated.
  • step S254 the evaluated stage is deemed a solution stage. Thereafter, in step S258, the stages are arranged along the die rail, and in step S256, the necessary left-right clearances for stage juxtapositioning are computed.
  • TE collision checking process referred to in each of steps S234, S238 and S246, may be performed as follows:
  • the tooling expert collision checking comprises mainly an intersection determination.
  • the intermediate part corresponding to the particular bend being evaluated in the search is formed, and is further converted to a B-rep (boundary representation) which is compatible with the NOODLES geometric modeler. Then, an intersection is performed utilizing the appropriate NOODLES function.
  • the number of faces of the part, as it changes shape throughout performance of the bend are monitored. For each of a plurality of discretized shapes of the part throughout performance of the bend, each of those shapes are intersected with the appropriate tools of the bending workstation during the performance of the bend. The resulting number of faces of the part, for each shape, is then counted. If the resulting number of faces, intersected with the tools, is greater than the expected number for that shape, then there has been a collision.
  • the above-described steps define a preferred algorithm for performing a tooling expert collision checking process.
  • the intermediate part before and after the bend may be modeled by a bounding box, and the basic solid intersection function provided by NOODLES may be utilized to determine if the tools intersect with the bonding box representation of the workpiece for the particular bend being evaluated during the search.
  • a process for determining the necessary left-right clearances for juxtapositioning the stages along the die rail as computed in step S256 of the process illustrated in Figs. 47A-47B.
  • the lateral limits of the part at the particular bend being evaluated are calculated based upon the amount by which the workpiece extends beyond a side edge of the solution tooling stage, and a largest lateral limit for each side of the stage is determined.
  • the stages arranged adjacent to the present solution stage are then appropriately spaced to have a gap between the adjacent side edges which is greater than or equal to the larger of the determined largest lateral limits of the adjacent side edges.
  • step S258 of the stage planning process shown in Figs. 47B-47C the present solution stage (corresponding to the presently evaluated bend) is placed in the middle of the die rail if it is the longest solution stage that has been evaluated so far in the search.
  • it is placed at the first or left position along the die rail. All middle gradations, from the second largest down, are respectively positioned from the third position to the last position along the die rail, the third position being positioned just to the right of the middle position, and the last position being the position furthest to the right.
  • a search algorithm could be used, such as A*, in order to come up with an appropriate stage layout that can accommodate co-linear bends, while minimizing the number of stages and the spacing between stages that are needed.
  • a significant cost to be taken into account by such a search algorithm is the total length of the die rail, the amount of space along the die rail a certain staging solution will occupy, and amount of space along the die rail remaining at the present level of the bend sequence (being generated by the bend sequence planner).
  • Figs. 48A-48C illustrate respective intermediate representations of a workpiece, being modeled in relation to the tooling during performance of a bend.
  • the workpiece In Fig. 48B, the workpiece is at a right position along the stage.
  • the bend line In each of Figs. 48A and 48B, the bend line is shorter than the length of the tooling stage.
  • Fig. 48C the workpiece centered along the tooling stage, where the bend line is slightly longer than the length of the tooling stage.
  • Fig. 49 illustrates a fine motion planning process, which may be performed in planning block P14 of the dialogue chart shown in Fig. 30.
  • a first step S260 of the process illustrated in Fig. 49 parameters are set and initialization steps are performed.
  • the 3D models of the tooling and the part are read, and various initialization functions are performed.
  • the goal parameters are set up based upon the tool and part geometry, and the desired clearance.
  • the portion of the part inside the bend line is rapidly analyzed, and a bounding box that surrounds the part is computed.
  • step S262 a determination is made as to whether or not a simple solution path is readily available, by testing if the top of the part can clear the bottom edge of the tooling punch, and if certain features of the part satisfy constraints imposed by the tool geometry and the die opening. If such a simple solution path is readily available, the process proceeds to step S264, where a fine motion plan is quickly generated. The process is then forwarded to step S270 where it returns to the tooling expert with the fine motion plan and the fine motion cost, which is equal to the amount of time that it takes to unload the part from the bend press.
  • step S266 a modified A* search is performed.
  • a plurality of feasible virtual configuration space nodes are generated and placed on the OPEN list with their respective costs.
  • the first level of the search includes several generated intelligent direction feasible VC (virtual configuration)-space nodes that were appended to the OPEN list.
  • VC virtual configuration
  • a node from the OPEN list is expanded, it is expanded to include several neighborhood nodes representative of locations in the general neighborhood of the parent node. Each expanded node is tested for feasibility by utilizing a geometric intersection test. If the test is positive (i.e., there is no collision by the use of a negative intersection function), the expanded node is appended to the OPEN list along with its cost.
  • the cost is an h cost which is set equal to the Euclidean distance from the expanded node to the goal.
  • the nodes on the OPEN list are continually expanded to lower levels in the search tree until the goal is reached or until the OPEN list becomes empty.
  • step S268 a determination is made as to whether or not the goal was reached. If the goal was reached, the fine motion planning process returns to the tooling expert with the fine motion costs and the fine motion plan in step S270. If the goal was not reached, the process proceeds to step S272, where the fine motion cost is set to infinity, and is sent to the tooling expert.
  • Fig. 50 illustrates an example process for determining the motion expert k and h costs, as indicated in planning box P21 of the dialogue chart shown in Fig. 31.
  • the k cost is calculated to be equal to a calculated robot travel time to take the part from a position at a stage of an immediate preceding bend to the stage location corresponding to the presently evaluated bend in the search, without regard to collisions.
  • the h cost is calculated to be equal to the product of the running average of the k cost values for the previous bends and the presently evaluated bend, and the sum of the number of remaining bends and twice the number of remaining predicted repos that will have to be performed before all of the bends in the bend sequence are completed.
  • a state-space search algorithm may be performed to form each of the steps along the path from one point to another in order to bring the workpiece throughout its various stages in the bend sequence.
  • collision checking may be performed. In order to perform this collision checking, the workpiece, the robot, and the bend press may each be modeled, and intersection tests may be performed using the appropriate NOODLES functions.
  • FIG. 51 illustrates a geometric model of a press brake 304, a workpiece bounding box 300, and a robot 302.
  • the workpiece is modeled by a bounding box 300.
  • the position of the robot 302 and the modeled part 300 is shown in three positions extending between a stage used for the final bend of the bend sequence to a position at the far right of the diagram which corresponds to a position ready for unloading by the loader/unloader.
  • Each module of planning system 71 utilizes geometric modeling functions in order to analyze the physical relationships between various components of the bending workstation and the workpiece as it is being moved and developed.
  • Such geometric modeling functions may include representing stock, intermediate, and final parts, checking for interferences during motion planning and assisting in selecting robot grip positions.
  • needed geometry information may be provided to assist the sub-planners in determining punch geometry selection, tool placement, loader/unloader suction cup 31 placement, and interpretation of sensing signals.
  • Simplified geometric representations may be provided for fast computations (e.g., bounding boxes, convex hulls, and 2D cross-sections), which may be needed to perform geometric-based reasoning methods (e.g., oct-tree representations, and configuration spaces).
  • a geometric database of physical components may be provided which includes both symbolic descriptions (e.g., labeled features) along with actual geometry data of physical components.
  • Other geometric modeling functions may be provided, although they are not specifically enumerated herein.
  • NOODLES is utilized to perform many of the noted modeling functions.
  • NOODLES includes a large package of geometric routines and is accessible to C/C+/C++ source code.
  • NOODLES is capable of handling non-manifold geometry (e.g., 0D, 1D, 2D, 3D, etc.) with the same routines, and has a hierarchal structure which can be used to build geometry libraries and to store various types of information regarding features of parts.
  • a modeling mechanism may be provided for modeling both upper and lower surfaces (i.e., the thickness) of each sheet metal workpiece throughout one or more of the design, planning, and execution phases of the bending process. It may be useful to have such a complete thickness representation in the workpiece for certain aspects of the system. For example, holding expert 82 may benefit from the added knowledge of knowing both the upper and lower surfaces of the workpiece, and motion expert 84 may be able to better plan for and control fine motion of the work piece when it is close to the die and punch tool before and after a bending operation.
  • an upper/lower surface modeling mechanism (not shown) performs a thickness transformation between a flat representation 114 and a representation with thickness 116, shown at the right of Fig. 10.
  • the representation with thickness 116 comprises two flat representations juxtaposed one on the other.
  • Fig. 11 illustrates an overlapped flange 118 modeled as a flat representation 114 at the left of Fig. 11, and transformed to a representation with thickness (i.e., a solid model).
  • Solid model 116 is shown to be equal to an upper surface representation 120 together with a lower surface representation 122.
  • Upper surface representation 120 is shown in solid lines, and lower surface representation 122 is shown in dotted lines.
  • Fig. 12 represents an exemplary tree structure which may be utilized to model the design representation of a sheet metal workpiece 16.
  • a plurality of shapes 126 are indicated corresponding to workpiece 16.
  • several faces 128 are defined, and for each face, several edges 130 are defined.
  • a plurality of vertices 132 are indicated.
  • a 2D (i.e., stock part) representation 134 may be maintained, along with a 3D (i.e., final part) representation 136 and an intermediate representation 138.
  • a thickness transformation may be performed, as represented by arrow 140, resulting in upper and lower surface representations 142, 144, which each have a tree structure similar to that illustrated above the line in Fig. 12.
  • Figs. 17A-17B and 18A-18B illustrate several different types of geometric libraries which may be provided in order to aid in the performance of geometric modeling of the system.
  • FEL a query-based language
  • FEL was originally developed by David Bourne in 1988, and has since been further refined.
  • Feature Exchange Language Programmer's Guide David Alan Bourne, Duane T. Williams (January 14, 1994); "Using the Feature Exchange Language in the next Generation Controller,” David Alan Bourne, Duane T. Williams, CMU-RI-TR-90-19; and "The Operational Feature Exchange Language,” David Alan Bourne, Jeff Baird, Paul Erion, and Duane T. Williams, CMU-RI-TR-90-06.
  • Fig. 19 illustrates an exemplary FEL planning message 145 which is being sent from bend sequence planner 72, as indicated by expression 146, to motion expert 84, as indicated by expression 148.
  • FEL planning message 145 comprises a query command sent from bend sequence planner 72 to motion expert 84, which provides preliminary information to motion expert 84 so that it may satisfy the query.
  • An initial parameter setting portion 150 of message 145 is provided immediately after a main verb/command "get” 152, and includes expressions "type message” 147, "from planning” 146, "to moving” 148, and "state request” 149.
  • the expression "type cost” is provided immediately after setting portion 150, and signifies that a request is being made for the motion expert to tell the planner how much a particular operation will cost.
  • the next expression "bends " 156 queries how expensive it will be to perform bend number 3, after having done bend number 6.
  • the numbers 7 and 1 represent a face of the workpiece that will be inserted into the die space of the bending workstation for bends 6 and 3, respectively.
  • a next expression "average_cost 2.321” 158 informs the motion expert that this is the average cost (k-cost) for motion per bend for the bends that have previously been done based upon cost values previously assigned by the motion expert. In this case, the average cost is 2.321 seconds per bend previously performed.
  • a next expression “flange-_before_bend” 160 indicates the height (in millimeters) of the tallest flange of concern (indicated in Fig. 18A as 11 millimeters) to be used by the motion expert to make clearance determinations.
  • Expression “flange_after_bend” 162 similarly indicates the height (in millimeters) of the tallest flange of concern which will exist after the bend is performed (indicated in Fig. 18 as 17.5 millimeters).
  • a next expression "robot_loc" 164 informs the motion expert where the part is by specifying the location of the robot (as it was left upon completion of the previous bend).
  • a last expression in the planning message 145, "bendmap" 166 indicates the respective tool stages for the previous bend and presently proposed bend and where the workpiece should be with respect to the stage for each bend.
  • the first value 168 represents that the location information is given for bend number 6, and a second value 170 indicates the stage at which bend number 6 was performed, which in this case is stage number 1.
  • Several coordinates are listed to the right of the first and second values 168, 170.
  • the first coordinate value "257.” represents the position of the left edge of the part with respect to the left edge of the stage
  • the second coordinate value "-257” represents the position of the left edge of the part with respect to the stage
  • the value "350.7” represents the position of the right edge of the part with respect to the stage.
  • the final value "320.” represents the position of the stage along the die rail with respect to the left edge of the die rail.
  • the planning message 145 forwards all the information which the motion expert will need in order for it to generate a subplan for moving the workpiece from an initial position (where it is left after performance of a preceding bend) to a position ready for a proposed next bend.
  • a significant feature of the query-based interface structure between the planner and its various sub-planners (experts) is that when the planner forwards a query to an expert, it informs the expert of all background information that the expert will need to respond to the query. Thus, the experts need not save information, but can simply respond to the bend sequence planner and return all related information for the bend sequence planner to save.
  • each module including bend sequence planner 72, and experts 80, 82, and 84 is sent a command to read its startup configuration file.
  • An example of such a command could be as follows:
  • bend sequence planner 72 can use any specified number of experts, e.g., using a command such as the following:
  • bend sequence planner 72 may start the planning process.
  • bend sequence planner 72 in participating in a dialogue with the other modules of the system, including the experts.
  • FEL MODULE DIALOG COMMANDS SEARCH COMMANDS Finalize - collect final plan info from each module Get - get cost information (and other data) for a bend Plan - initialize a module for planning a part USER COMMANDS Quit - cleanup and exit a module Read - read files for planning Set - set various module options Show - show various module data to user
  • bend sequence planner 72 for execution by sequencer 77.
  • FEL SEQUENCER COMMANDS Print Messages - print messages for BM100 operator for setup
  • Programs download programs to NC9R press controller and backgage controller Startup - initialize state of press and robot Get - acquire part from various steps of the process Put - load part into various steps of the process Move - move the robot through a series of points Bend - initiate bend sequence (backgage and bending)
  • the "read” command may be used to instruct a module to read certain files needed for planning, the files being representative of the design to be produced, and to configure itself in accordance with the design.
  • various module functions may be set, e.g., how to display information, how to interface with other modules, and so on.
  • the "show” user command may be utilized to show various module data to the user, e.g., the various nodes of the A* algorithm which represents the various costs or different bends within the proposed bend sequence.
  • a CAD system 74 performs several functions relating to part design and part modeling for planning system 71.
  • CAD system 74 allows a user to form a design of a given workpiece by working with simplified, primitive components (in either 2D or 3D form) on a graphic interface, each primitive component having certain desired dimensions which may be input by the user, in order to design the workpiece.
  • the user may then utilize a user interface with CAD system 74 to connect the primitive components and, in addition, to remove portions, such as holes, slots, etc., from the connected primitive components.
  • CAD system 74 may then perform feature labeling functions including labeling several geometric features of the workpiece, such features having a particular significance in the context of sheet metal bending.
  • CAD system 74 may also build a bend graph which associates various bend-related information with the geometric design of the workpiece. CAD system 74 thereby forms an output file which includes geometric, topological, and bend-related feature information (including a list of labeled features and a bend graph). All of this information is then placed into an output shape file which will form the basis of communication with other modules of planning system 71.
  • a part modeler may be provided to form an interface between the design system's output shape file and the various expert modules 80, 82, and 84 (and 85) along with bend sequence planner 72.
  • a part modeler may be provided which performs various conversions on the data provided in the output shape file in order to form developed part data structures which can be used for geometric modeling purposes by each of the modules of planning system 71.
  • Part modeler may be implemented in the form of a library which is accessible to each of the modules in planning system 71, which may be utilized to manipulate the information in the developed part data structures and/or undeveloped data structures provided in the output shape file, in order for the various modules to utilize the information provided therein to serve any particular purpose that they may be addressing at a particular point in time.
  • Fig. 13A illustrates a functional block diagram of a design system 311 which may be provided to perform the functions of CAD system 74 of the illustrated embodiment.
  • Design system 311 performs several design-related functions which may be implemented in the form of function modules as illustrated in Fig. 13A.
  • Each function module may be implemented by a particular function provided in a library of functions comprised by the design system.
  • the functions shown in Fig. 13A include a user interface 312, file I/O 314, view 316, simulation 318, shape defining 320, hole defining 322, editing 324, and feature labeling 328.
  • Each of these functions may be controlled by a design system control module 326.
  • bend graph module 330 and bend deduction module 332 are each connected to feature labeling module 328.
  • Fig. 13A Each of the functions are illustrated in Fig. 13A in the form of function modules. However, it is not necessary that each of these functions be separated into separate modules in the specific manner as illustrated. In the alternative, an overall program or hardware system may be provided which allows each of these functions to be performed without having any specific interface with other functions of the design system. For example, one complete routine may be provided within a processor of a computer to implement each and every one of the functions of the overall design system, without removing several of the general benefits provided by the design system disclosed herein.
  • the file I/O module 314 performs functions such as reading, writing, printing, and performing data exchanges between modules.
  • the view function module 316 performs functions such as zooming in/out, and panning during display of the part on a graphic interface.
  • the shape module 320 is provided to allow a user to specify particular shapes, including rectangular shapes, angles, a Zee, a box, a hat, and so on, which may be put together to form a particular workpiece design.
  • Hole module 322 is provided for the user to specify various type of cavities to be provided in the workpiece, such as cutouts, holes, slots, notches and so on, to further allow the user to design the workpiece in a manner similar to that provided by shape module 320.
  • Edit module 324 is provided to allow the user to perform various editing functions such as a fillit function, a chamfer function, and changing the workpiece material type and/or thickness.
  • Simulation module 318 is provided so that the user can simulate bending and unfolding of various bends on the workpiece, thus to get a visual representation of such bends on the graphic interface to be utilized by the design system.
  • Feature labeling module 328 is provided to automatically assign feature labels which pertain to sheet metal bending, and which will thus be useful to the planning system 71 illustrated herein in forming or generating a bend sequence plan with the use of such feature labels.
  • Feature labeling module 328 may generate feature-related information such as corners, setbacks, form features (e.g., dimples, louvers), holes, large radius bend, etc.
  • feature labeling module 328 may be designed so that it directs a bend graph module 330 to form a bend graph which includes information organized in a certain way to relate the geographic and topological information to the various bends to be performed on the 2D workpiece to form the desired 3D finished workpiece.
  • feature labeling module 328 may be designed so that it directs the performance of bend deduction calculations by a bend deduction module 332. The resulting bend deduction information may then be placed within a bend graph listing provided by bend graph module 330.
  • a part modeler should be provided, and may be provided in the form of a library of functions accessible to the various modules in order to interface between the design system's output shape files and the various modules within planning system 71.
  • Fig. 13B illustrates a part modeling system 333 for performing this function.
  • Part modeler 333 includes two main function modules: a B-REP rearrangement module 336 and an intermediate shape conversion module 342.
  • the B-REP rearrangement module 336 converts an undeveloped part data structure 334 to either or both of a developed 3D part data structure (in B-REP) 338 and a developed 2D part data structure (in B-REP) 340.
  • Intermediate shape conversion module 342 converts the developed 2D part data structure (in B-REP 340) to a developed intermediate part data structure (in B-REP) 344.
  • the undeveloped part data structure 334 defines a geometric/topological data structure that does not take into account bend deduction and that forms part of the output shape file produced by CAD system 74.
  • a developed part data structure such as developed 3D part data structure 338 and developed 2D part data structure 340, includes a modified representation of the part that takes into account bend deduction.
  • the noted developed part data structures are further converted to be in the form of a boundary representation (B-rep) model.
  • the data structure which resides in the shape output file produced by the CAD system may be designed to include a shape header which includes part information, followed by a plurality of shapes in a linked list, the linked list ending with a null.
  • topological and geometric information may be provided for both a 3D and a 2D representation of the part.
  • the structure of the shape may include a list of information including the shape type, shape identification, a face list, an edge list, a 3D vertices list, and a 2D vertices list.
  • Each face may have its own structure, which may include a list of information including a face identification, the number of vertices of the face, a vertices list for the vertices of the face, and a face normal vector.
  • a structure For each edge, a structure may be provided which includes information such as the edge identification, the edge type, the bent line type, and the vertices index number for that particular edge. For each vertex, information may be provided including the vertices identification, vertices coordinate, 2D coordinates, 3D coordinates and intermediate coordinates. Further information regarding the details of data structures and the illustrated CAD system in general are provided in an ME report dated May, 1992 entitled “A Parallel Design System for Sheet Metal Parts" presented by Cheng-Hua Wang at the Mechanical Engineering Department, Carnegie Mellon University, Pittsburgh, Pennsylvania
  • the CAD system preferably employs a concurrent "parallel" representation of both the 3D and the 2D versions of the part as it is being designed, and such representations are maintained once the part is finally designed for use by planning system 71.
  • Figs. 13C and 13D are provided.
  • One of the benefits of having a concurrent and parallel design system is that such a system resolves ambiguities which may otherwise occur in the design process.
  • a 2D part 346a is illustrated in Fig. 12C and a 3D part 346b is shown in Fig. 12D.
  • Another benefit associated with such a concurrent design approach is that it may be easier to make modifications to one representation (e.g., the 2D representation) instead of the other for a particular type of modification, e.g., adding an inner tab to the part.
  • one representation e.g., the 2D representation
  • Figs. 14A-14E illustrate a design system graphical user interface 348, with its display changing throughout the process of designing a certain desired part.
  • graphical user interface 348 includes a key pad 350, a parameters window 352, a primitive shape 3D window 354, a primitive shape 2D window 356, a model 3D window 358 and a model 2D window 360.
  • Fig. 14A shows the first introduced primitive shape provided on a graphical interface 348 in order to produce the desired workpiece as shown in Fig. 14E.
  • the first primitive shape is a box.
  • the parameters of the box may be specified with the use of key pad 350 and are illustrated in parameters window 352 to have a base which is 100 x 100 (indicated by parameters P[1] and P[2]), and a height equal to 20 (indicated by parameter [3]).
  • the 3D version of the primitive shape is illustrated in primitive shape 3D window 354, and the 2D shape of the primitive shape is illustrated in primitive shape 2D window 356. Since this is the first primitive shape being provided for the part design, model 3D window 358 is identical to primitive shape 3D window 354, and model 2D window 360 is identical to primitive shape 2D window 356.
  • Fig. 14B illustrates the next shape to be added which is a rectangle having a length of 100 (indicated by parameter [1]), and a width of 15 (indicated by parameter [2]).
  • the next primitive shape being added to design the part is another rectangle having the same parameters as the rectangle of Fig. 14B.
  • the next primitive shapes are added to the workpiece as shown in Figs. 14C, 14D and 14E.
  • Parameter P[1] corresponds to the X dimension
  • parameter P[2] corresponds to the Y dimension
  • parameter P[3] corresponds to the Z dimension of the primitive shape being added.
  • Figs. 15A-15C are provided to illustrate bend deduction, and the manner in which it relates to the 3D and 2D dimensions of flanges of a workpiece.
  • a workpiece 362 has a thickness t
  • the flanges of the workpiece 362 are desired to have lengths a and b
  • a calculation should be performed so that the flat 2D representation of the part, when bent along the appropriate bend line, will indeed form the flanges having appropriate dimensions a and b, taking into account the thickness t of the material, the material type, and the internal radius of the bend line (to the inside surface of the sheet metal).
  • the developed 2D representation 364 of workpiece 362 may be calculated by subtracting the appropriate bend deduction (BD) value from the overall dimension a + b.
  • BD bend deduction
  • Fig. 16 illustrates a graphic representation of a bend graph, the graphic representation being a 2D representation of the workpiece designed in the steps illustrated in Figs. 14A-14E.
  • the bend lines of the designed workpiece are labeled as bend lines B1, B2, ... B8, and each label comprises a bend line index.
  • Each bend line index is then assigned a bend sequence number which comprises an initialization value.
  • the bend sequence number indicates the order in the bend sequence in which the bend line will be bent, and is assigned for each bend line in accordance with the plan (i.e., the bend sequence) produced by the bend sequence planner of the illustrated planning system 71.
  • each bend line is assigned a bend angle. For example, in the bend graph illustrated in Fig. 16, an angle of -90.0° is given for bend B2, and a bend angle of 90.0° is given for B1.
  • the bend graph further comprises an indication of the various faces F1-F9 which are formed on the workpiece once the bends are performed.
  • Fig. 52 comprises a block diagram of the various software modules and their main interfacing components, such modules including planner 72, sequencer task 76, robot task 92, press and L/UL task 94 and backgage tasks96, speed control task 102, and collision detection task 100.
  • Planner 72 includes interfacing components such as an output queue 72a and input queue 72b.
  • the sequencer task 76 includes an input queue 76a, an output queue 76b, a task response queue 76c and a section corresponding to several task class member functions 76d.
  • Output queue 72a of planner 76 is connected to input queue 76a of sequencer task 76.
  • Output queue 76b of sequencer 76 is connected to input queue 72b of planner 72.
  • Robot task 92 includes an input queue 92a, an output queue 92b, and a portion corresponding to robot task functions 92c.
  • Press and L/UL task 94 includes an input queue 94a, an output queue 94b, and a portion corresponding to press task functions and L/UL task functions 94c.
  • Backgage task 96 includes an input queue 96a, an output queue 96b, and a portion corresponding to backgage task functions 96c.
  • Each of input queues 92a, 94a, and 96a is connected to input queue 76a of sequencer task 76.
  • Each of output queues 92b, 94b, and 96b is connected to task response queue 76c of sequencer task 76.
  • the controller software structure shown in Fig. 52 is representative only of an example of the inner connections between planner 72, sequencer task 76, and control system 75, the structure of each of the tasks, and how they are connected. It is within the scope of the invention disclosed herein to provide variations of a control system which performs the same essential controlling functions, without being implemented in the manner illustrated in Fig. 52.
  • Fig. 53 illustrates an example flow of the process performed by sequencer task 76 illustrated in Fig. 52.
  • the sequencer will obtain a new message from the FEL listing at input queue 76a.
  • the sequencer will parse the FEL sentence, and in step S284, the sequencer will create a data object for each task involved.
  • the appropriate data objects will be placed upon their appropriate task queues (e.g., on one or more of the input queues of robot task 92, press and L/UL task 94, and backgage task 96).
  • the sequencer checks the state of all tasks involved.
  • step S290 a determination is made as to whether all the tasks are finished. If not, the sequencer proceeds to step S292. If all the tasks have finished, the sequencer proceeds from step S290 to step S294 where appropriate cleanup operations are performed (e.g., destroying data objects and resetting flags).
  • step S292 a determination is made as to whether or not a time out has been exceeded. If not, the process returns to step S288. If the time out has been exceeded, the sequencer proceeds to step S293 where appropriate error recovery processing is performed. After the cleanup operations are performed in step S294, a determination is then made in step S296 as to whether the task exit signal has been set. If the task exit signal has been set, the process will then terminate. Otherwise, the process will return to step S280 where a new message will be acquired from the FEL input queue.
  • Fig. 54 is a flow chart of the overall bending process during execution of a single bend.
  • the robot places the part into the die space. Thereafter, the part is aligned in the X, Y and rotation directions. This alignment is part of the backgaging operation.
  • the press table is raised to the pinch point, i.e., the point at which the die contacts the workpiece, which in turn engages with the punch tool so that the workpiece is in a semi-stable state pinched between the die and tool punch.
  • the bend is executed with bend following (i.e., with the robot gripper maintaining its hold on the workpiece throughout the execution of the bend).
  • the press brake will be opened.
  • step S306 the part is unloaded from the die space. Once the part is unloaded, the bend is completed.
  • Fig. 55 illustrates the robot task 92 and the various functions that may be provided therein, including general motion functions and sensor-based motion functions.
  • the general motion functions may include a joint space move a cartesian move, and rotation about a point.
  • the sensor-based motion functions may include a guarded move, bend following, open loop bend, active damping, contact control, and compliant-part loading.
  • Compliant-part loading comprises loading a vibrating compliant-part into the die space of the proper timing so that the part fits in the die space and does not collide with the workstation.
  • Fig. 56 illustrates the press and L/UL (loader/unloader) task 94, and the various functions that may be provided within the task.
  • the functions that may be provided for controlling the press may include raise press, lower press, and bend.
  • the L/UL functions may include a load workpiece, release workpiece, grasp product, and unload product.
  • Fig. 57 illustrates the backgage task 96, and the various functions that may be provided therein.
  • the backgage task may include general motion functions and sensor-based motion functions.
  • One general motion function may include a move function.
  • the sensor-based motion functions may include a find part edge and a guarded move function.
  • the bend system illustrated herein may be provided with one or more mechanisms for learning from the results of the one or more initial runs of a plan, and for modifying the plan accordingly in order to improve the speed of operations and to also improve the quality of the resulting workpiece.
  • a sensor-based control mechanism may be provided for performing an operation, including moving a workpiece from one position to another.
  • the bending apparatus may use a sensor output to modify the movement of the workpiece, but measure the amount by which the movement of the workpiece is modified due to the sensor output. Then, by learning the amount by which the movement of the workpiece was modified, the operation may then be controlled, based upon what was learned, so that the workpiece is moved from one position to another without modifying the movement of the workpiece utilizing a sensor output.
  • Fig. 58 illustrates an example process for performing learning measurements and for modifying movement control during multiple executions of a generated bend sequence plan, where the movement of the workpiece from one position to another comprises droop compensation and backgaging in the X direction.
  • the sensor output comprises a measured amount of X offset and a measured amount of droop offset of the part.
  • the part is loaded for bending using droop sensing.
  • the amount of offset of the part i.e., the amount by which the part is drooping, is sensed and sent back to the planner (e.g., planner 72 illustrated in Figs. 5A and 6).
  • the part is side-gaged (gaged in the X direction) to obtain an X offset value.
  • the X offset value detected for this bend is sent back to the planner (or the process manager).
  • Backgaging is then performed to align the part in the Y direction and also to appropriately rotate the part so that it is in the appropriate yaw position.
  • the bend is then performed.
  • step S320 a determination is made as to whether or not there are more bends to be performed in the present bend sequence being executed. If so, the process returns to step S308, where steps S308-S318 are again performed to obtain values corresponding to that next bend. If all of the bends have been completed, the process proceeds from step S320 to step S322, at which point the finished part is unloaded, and a new workpiece is loaded with the loader/unloader. Then in step S324, the part is loaded for bending using the measured droop offset and measured X offset values that were previously determined and forwarded to the planner. By using such values, the bending apparatus can position the workpiece without performing sensor-based control (or at least with a simplified sensor-based control method) while positioning the workpiece.
  • step S326 backgaging is performed to align the part in the Y and rotation (yaw) directions.
  • the bend is then performed in step S328, and a determination is then made in step S330 as to whether more bends in the bend sequence still have to be performed. If all the bends have been performed, the process proceeds to step S332, at which point a determination is made as to whether more parts are to be made. If more parts are to be made, the process returns to step S322.
  • the offset values Due to the repeatability of a typical bending workstation, such as the Amada BM100 bending workstation, the offset values only need to be determined by performing one or a few execution runs of the system. Once the offset values are determined, the offset values may be used for future batch runs of the system, and should be considered dependable for many runs. Accordingly, the process in Fig. 58 is illustrated as returning from step S332 to step S322 for each new workpiece to be formed, rather than returning all the way back to step S308 for obtaining new offset values.
  • the present invention is described as being directed to methods and subsystems provided in an intelligent design, planning and manufacturing system for producing materials such as bent sheet metal parts.
  • the present invention may be further utilized for performing such functions as costing (i.e., determining how much it will cost to develop certain types of parts with a given sheet metal bending work station), scheduling (e.g., determining how much time it will take to perform to manufacture various parts with a given sheet metal bending work station) and part design and assembly.
  • the planning system 71 of the present invention e.g., as disclosed in Fig. 5A
  • the planning system 71 of the present invention is capable of generating a complete sequence of bends and bend-related operations which will be needed to form a given part.
  • the generated sequence of operations may be accompanied by a complete plan which specifies all steps needed to execute the bend sequence in a proper order by the sheet metal bending work station.
  • the planning system 71 through use of experts/subplanners, will determine the consequences of performing each bend and other accompanying operations within the bend sequence. Accordingly, without actually executing the resulting plan generated by planning system 71, planning system 71 will have information as to what the likely amount of time it will take to perform all of the necessary operations to manufacture the part with the sheet metal bending work station.
  • the planning system 81 will be able to further confirm whether or not the sheet metal bending work stations and available tooling are capable of forming a particular designed part.
  • planning system 71 can determine the resulting costs, and such information may be utilized to evaluate the cost of producing a given set of parts that form a desired assembly.
  • planning system 71 will be able to determine factory scheduling with its information regarding the time needed to complete various operations of the plan. In addition, by knowing the limitations of producing a particular part, the amount of time it would take to produce the part, and the costs, it will be possible to utilize such information to generate alternative part designs which may result in less cost and less time needed for production of the part.
  • planning system 71 has been described specifically as comprising a plurality of experts, with each expert being implement in the form of a module which is separate from bend sequence planning module 72, planning system 71 may be implemented without being separated into modules.
  • planning system 71 may be implemented as one overall operations planning module.
  • the language utilized to communicate between the respective modules may be a language other than FEL.
  • each module i.e., bend sequence planner 72 and subplanners 80, 82, 84 and 85
  • each module may be implemented on a different computer/processor.

Claims (27)

  1. Dispositif de pliage comprenant un moyen de commande configuré pour générer un plan comprenant une séquence d'opérations que le dispositif de pliage (10) doit exécuter pour plier des pièces de travail (16) comprenant des feuilles d'une matière malléable, ladite séquence d'opérations comprenant un ensemble de N plis pour former une pièce de travail finie (16) à partir d'une feuille brute d'une matière malléable, ledit moyen de commande comprenant :
    des moyens de proposition pour proposer, pour une mième opération à l'intérieur de la séquence d'opérations, une pluralité d'opérations proposées comprenant une pluralité de plis proposés devant être exécutés par ledit dispositif de pliage (10);
    des moyens de sous-plan pour fournir un sous-plan proposé qui accompagne chaque pli proposé; et
    des moyens de génération pour générer un plan comprenant une séquence de plis allant d'un premier pli à un Nième pli, en choisissant chaque pli dans la séquence d'opérations sur la base des plis proposés et du sous-plan proposé qui accompagne chaque pli proposé, dans lequel les moyens de commande comprennent en outre des moyens d'estimation d'un coût à associer à chaque pli proposé.
  2. Dispositif de pliage selon la revendication 1, caractérisé lesdits moyens de génération génèrent un plan comprenant une séquence de plis allant d'un premier pli à un Nième pli, en choisissant chaque pli dans la séquence d'opérations sur la base des plis proposés, du sous-plan proposé qui accompagne chaque pli proposé et du coût estimé associé à chaque pli proposé.
  3. Dispositif de pliage selon la revendication 2, caractérisé en ce que les coûts estimés associés à un nième pli dans la séquence de N plis comprennent un coût k calculé sur la base d'une quantité de temps estimée que prendra le dispositif de pliage (10) pour accomplir une ou plusieurs opération(s) de pliage.
  4. Dispositif de pliage selon la revendication 2 ou 3, caractérisé en ce que le coût estimé associé à un nième pli dans la séquence de N plis comprend un coût h calculé sur la base d'une quantité de temps totale estimée que prendra le dispositif de pliage (10) pour accomplir une ou plusieurs opération(s) de chacun du reste des plis dans la séquence de plis qui suit le nième pli.
  5. Dispositif de pliage selon la revendication 3 ou 4, caractérisé en ce qu'une ou plusieurs opération(s) d'un pli donné comprend/comprennent le déplacement de la pièce de travail (16) à partir d'une position d'étape de formation d'un pli précédent jusqu'à une position d'étape de formation du pli donné.
  6. Dispositif de pliage selon au moins une des revendications 3 à 5, caractérisé en ce que ladite une ou lesdites plusieurs opération(s) d'un pli donné comprend/comprennent l'installation, lors du réglage du dispositif de pliage (10), d'une étape d'usinage supplémentaire qui sera nécessaire pour exécuter le pli donné.
  7. Dispositif de pliage selon au moins une des revendications 3 à 6, caractérisé en ce que ladite une ou lesdites plusieurs opération(s) d'un pli donné comprend/comprennent le repositionnement des pinces (14) supportées sur la pièce de travail (16) avant l'exécution du pli donné.
  8. Dispositif de pliage selon au moins une des revendications 2 à 7, caractérisé en ce que lesdits moyens de proposition et lesdits moyens de génération comprennent collectivement un module de planification de séquence de plis (72), et dans lequel, en outre, lesdits moyens de sous-plan et lesdits moyens d'estimation comprennent collectivement une pluralité de modules experts (80, 82, 84, 85).
  9. Dispositif selon la revendication 8, caractérisé en ce que lesdits modules experts (80, 82, 84, 85) comprennent chacun des moyens pour utiliser lesdits moyens de sous-plan et lesdits moyens d'estimation lorsque lesdits moyens de proposition proposent une opération proposée à exécuter comme ladite mième opération à l'intérieur de la séquence d'opérations.
  10. Dispositif de pliage selon au moins une des revendications 8 ou 9, caractérisé en ce que ladite pluralité de modules experts comprend un module expert de support (82) comprenant des moyens pour utiliser lesdits moyens de sous-plan pour fournir un sous-plan de support proposé comprenant des informations relatives à une position sur la pièce de travail (16) à laquelle les pinces (14) peuvent saisir la pièce de travail (16) tout en exécutant les plis de la séquence de plis.
  11. Dispositif de pliage selon au moins une des revendications 8 à 10, caractérisé en ce que ladite pluralité de modules experts comprend un module expert de support (82) comprenant des moyens pour utiliser lesdits moyens d'estimation pour estimer un coût de support calculé en se basant sur le fait de savoir si des pinces (14) supportées sur la pièce de travail (16) doivent être repositionnées avant l'exécution d'un pli donné.
  12. Dispositif de pliage selon au moins une des revendications 8 à 11, caractérisé en ce que ladite pluralité de modules experts comprend un module expert d'usinage (80) comprenant des moyens pour utiliser lesdits moyens de sous-plan pour fournir un sous-plan d'usinage proposé comprenant des informations relatives à une position le long d'une étape d'usinage à laquelle la pièce de travail (16) sera chargée dans le dispositif de pliage (10) en vue d'exécuter un pli donné.
  13. Dispositif de pliage selon au moins une des revendications 8 à 12, caractérisé en ce que ladite pluralité de modules experts comprend un module expert d'usinage (80) comprenant des moyens pour utiliser lesdits moyens d'estimation pour estimer un coût sur la base d'une quantité de temps nécessaire pour installer, lors du réglage du dispositif de pliage (10), une étape d'usinage supplémentaire nécessaire pour exécuter un pli donné.
  14. Dispositif de pliage selon au moins une des revendications 8 à 13, caractérisé en ce que ladite pluralité de modules experts comprend un module expert de mouvement (84) comprenant des moyens pour utiliser lesdits moyens d'estimation pour estimer un coût sur la base d'un temps de déplacement calculé pour déplacer la pièce de travail (16) d'une position d'étape d'usinage d'un pli à une position d'étape d'usinage d'un pli suivant.
  15. Dispositif de pliage selon au moins une des revendications 8 à 14, caractérisé en ce que ledit module de planification de séquence de plis comprend des moyens pour interroger chacun desdits modules experts pour un sous-plan et pour des coûts estimés, et dans lequel chacun desdits modules experts comprend des moyens pour répondre à une interrogation en renvoyant une liste de sauvegarde audit module de planification de séquence de plis, ladite liste de sauvegarde comprenant une liste de noms d'attributs ainsi que des valeurs correspondant respectivement aux attributs à sauvegarder par ledit module de planification de séquence de plis.
  16. Dispositif de pliage selon au moins une des revendications 1 à 15, caractérisé en ce qu'il comprend en outre des moyens de classement par ordre de priorité pour classer par ordre de priorité des plis proposés selon les caractéristiques heuristiques des plis déterminées sur la base de la géométrie de la pièce de travail (16).
  17. Dispositif de pliage selon la revendication 16, caractérisé en ce que lesdits moyens de génération génèrent un plan comprenant une séquence de plis allant d'un premier pli à un Nième pli, en choisissant chaque pli dans la séquence d'opérations sur la base des plis proposés classés par ordre de priorité et du sous-plan proposé qui accompagne chaque pli proposé.
  18. Dispositif de pliage selon au moins une des revendications 2 à 17, caractérisé en ce qu'il comprend en outre des moyens de classement par ordre de priorité pour classer par ordre de priorité des plis proposés selon les caractéristiques heuristiques de plis déterminées sur la base de la géométrie de la pièce de travail (16).
  19. Dispositif de pliage selon la revendication 18, caractérisé en ce que lesdits moyens de classement par ordre de priorité comprennent des moyens pour abaisser un coût estimé d'un pli présentant une priorité élevée et pour augmenter un coût estimé pour un pli présentant une priorité basse.
  20. Dispositif de pliage selon au moins une des revendications 1 à 19, caractérisé en ce qu'il comprend en outre des moyens de détermination pour déterminer le temps nécessaire pour produire, et la faisabilité de production d'une ou de plusieurs pièces avec le dispositif de pliage (10) sur la base du plan généré.
  21. Dispositif de pliage selon la revendication 20, caractérisé en ce qu'il comprend en outre des moyens pour exécuter des calculs du coût de production d'un lot donné des pièces sur la base du temps déterminé par lesdits moyens de détermination.
  22. Dispositif de pliage selon la revendication 20 ou 21, caractérisé en ce qu'il comprend en outre des moyens pour re-concevoir la pièce sur la base des déterminations de temps et de faisabilité réalisées par lesdits moyens de détermination.
  23. Dispositif de pliage selon au moins une des revendications 20 à 22, caractérisé en ce qu'il comprend en outre des moyens pour programmer la fabrication par le dispositif de pliage en fonction de la quantité de temps déterminée pour produire une ou plusieurs pièce(s).
  24. Dispositif de pliage selon au moins une des revendications 1 à 23, caractérisé en ce que lesdits moyens de proposition comprennent des moyens pour proposer des plis parmi l'ensemble complet de N plis encore restants.
  25. Dispositif de pliage selon au moins une des revendications 1 à 24, caractérisé en ce que lesdits moyens de proposition comprennent des moyens pour proposer des plis parmi l'ensemble complet de N plis encore restants moins les plis bloqués en raison de certaines contraintes.
  26. Dispositif de pliage selon au moins une des revendications 1 à 25, caractérisé en ce lesdits moyens de proposition comprennent des moyens pour proposer, pour une mième opération, un repositionnement des pinces (14) supportées sur la pièce de travail (16).
  27. Dispositif de pliage selon au moins une des revendications 1 à 26, caractérisé en ce qu'il comprend en outre des moyens pour représenter la mième opération à l'intérieur de la séquence d'opérations comme le mième niveau d'un arbre de recherche.
EP02002809A 1994-11-09 1995-11-09 Dispositif de pliage comprenant un système de contrôle servant à générer et à exécuter un plan de pliage de tôles métalliques Expired - Lifetime EP1253496B1 (fr)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP06009075A EP1684140B1 (fr) 1994-11-09 1995-11-09 Dispositif de flexion incluant un moyen de contrôle et procédé correspondant
EP06002177A EP1657607B1 (fr) 1994-11-09 1995-11-09 Appareil de pliage d'une feuille métallique comprenant un dispositif de contrôle
EP06009079A EP1681607B1 (fr) 1994-11-09 1995-11-09 Dispositif de flexion incluant des moyens pour effectuer les opérations de réglage

Applications Claiming Priority (5)

Application Number Priority Date Filing Date Title
US33811394A 1994-11-09 1994-11-09
US338113 1994-11-09
US386369 1995-02-09
US08/386,369 US5969973A (en) 1994-11-09 1995-02-09 Intelligent system for generating and executing a sheet metal bending plan
EP95936762A EP0744046B1 (fr) 1994-11-09 1995-11-09 Systeme intelligent servant a generer et a executer un plan de pliage de toles metalliques

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
EP95936762A Division EP0744046B1 (fr) 1994-11-09 1995-11-09 Systeme intelligent servant a generer et a executer un plan de pliage de toles metalliques

Related Child Applications (3)

Application Number Title Priority Date Filing Date
EP06002177A Division EP1657607B1 (fr) 1994-11-09 1995-11-09 Appareil de pliage d'une feuille métallique comprenant un dispositif de contrôle
EP06009075A Division EP1684140B1 (fr) 1994-11-09 1995-11-09 Dispositif de flexion incluant un moyen de contrôle et procédé correspondant
EP06009079A Division EP1681607B1 (fr) 1994-11-09 1995-11-09 Dispositif de flexion incluant des moyens pour effectuer les opérations de réglage

Publications (2)

Publication Number Publication Date
EP1253496A1 EP1253496A1 (fr) 2002-10-30
EP1253496B1 true EP1253496B1 (fr) 2006-05-03

Family

ID=23323468

Family Applications (4)

Application Number Title Priority Date Filing Date
EP06002177A Expired - Lifetime EP1657607B1 (fr) 1994-11-09 1995-11-09 Appareil de pliage d'une feuille métallique comprenant un dispositif de contrôle
EP06009079A Expired - Lifetime EP1681607B1 (fr) 1994-11-09 1995-11-09 Dispositif de flexion incluant des moyens pour effectuer les opérations de réglage
EP06009075A Expired - Lifetime EP1684140B1 (fr) 1994-11-09 1995-11-09 Dispositif de flexion incluant un moyen de contrôle et procédé correspondant
EP02002809A Expired - Lifetime EP1253496B1 (fr) 1994-11-09 1995-11-09 Dispositif de pliage comprenant un système de contrôle servant à générer et à exécuter un plan de pliage de tôles métalliques

Family Applications Before (3)

Application Number Title Priority Date Filing Date
EP06002177A Expired - Lifetime EP1657607B1 (fr) 1994-11-09 1995-11-09 Appareil de pliage d'une feuille métallique comprenant un dispositif de contrôle
EP06009079A Expired - Lifetime EP1681607B1 (fr) 1994-11-09 1995-11-09 Dispositif de flexion incluant des moyens pour effectuer les opérations de réglage
EP06009075A Expired - Lifetime EP1684140B1 (fr) 1994-11-09 1995-11-09 Dispositif de flexion incluant un moyen de contrôle et procédé correspondant

Country Status (3)

Country Link
US (1) US5969973A (fr)
EP (4) EP1657607B1 (fr)
DE (3) DE69535946D1 (fr)

Families Citing this family (72)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE69529607T2 (de) 1994-11-09 2003-09-18 Amada Co Intelligentes system zur herstellung und ausführung eines metallplattenbiegeplans
US5835684A (en) * 1994-11-09 1998-11-10 Amada Company, Ltd. Method for planning/controlling robot motion
US5761940A (en) * 1994-11-09 1998-06-09 Amada Company, Ltd. Methods and apparatuses for backgaging and sensor-based control of bending operations
US5822207A (en) * 1996-05-06 1998-10-13 Amadasoft America, Inc. Apparatus and method for integrating intelligent manufacturing system with expert sheet metal planning and bending system
US5828575A (en) 1996-05-06 1998-10-27 Amadasoft America, Inc. Apparatus and method for managing and distributing design and manufacturing information throughout a sheet metal production facility
US5971589A (en) 1996-05-06 1999-10-26 Amadasoft America, Inc. Apparatus and method for managing and distributing design and manufacturing information throughout a sheet metal production facility
US6662061B1 (en) 1997-02-07 2003-12-09 Peter G. Brown System and method for simulation and modeling of batch process manufacturing facilities using process time lines
CA2285351A1 (fr) 1997-02-07 1998-08-27 Peter G. Brown Systeme et procede servant a simuler et a modeliser des installations de preparation par lots de produits biopharmaceutiques
US6338000B1 (en) * 1997-03-24 2002-01-08 Honda Giken Kogyo Kabushiki Kaisha Method of generating shape data method of verifying shape data
US6983229B2 (en) * 1997-06-20 2006-01-03 Brown Peter G Method for scheduling solution preparation in biopharmaceutical batch process manufacturing
US7043414B2 (en) * 1997-06-20 2006-05-09 Brown Peter G System and method for simulating, modeling and scheduling of solution preparation in batch process manufacturing facilities
US6311093B1 (en) 1997-06-20 2001-10-30 Peter G. Brown System and method for simulation, modeling and scheduling of equipment maintenance and calibration in biopharmaceutical batch process manufacturing facilities
IT1294998B1 (it) * 1997-09-09 1999-04-27 Salvagnini Italia Spa Procedimento per la generazione automatica di una sequenza di comandi per macchina piegatrice di fogli di lamiera
US6233538B1 (en) * 1997-09-11 2001-05-15 Amada America, Inc. Apparatus and method for multi-purpose setup planning for sheet metal bending operations
US6289255B1 (en) * 1997-11-25 2001-09-11 Voyan Technology Method for computer-aided design of a product or process
US6826449B1 (en) * 1997-12-30 2004-11-30 Ali F. Abu-Taha Method for producing natural motions
US6292715B1 (en) * 1998-10-27 2001-09-18 Perry Investments, Inc. Robotic process planning method and apparatus using templates
DE60044022D1 (de) * 1999-01-13 2010-04-29 Amada Co Ltd Biegepresssystem
US6377871B1 (en) * 1999-10-26 2002-04-23 Motoman, Inc. Integrated die cast
US6684116B1 (en) * 2000-01-31 2004-01-27 Autodesk, Inc. Method and apparatus for incorporating features in sheet bodies
KR100331632B1 (ko) * 2000-04-14 2002-04-09 이계안 트랜스퍼 프레스용 핑거 제작 방법
US6545680B1 (en) * 2000-06-15 2003-04-08 Dessault Systemes S.A. Dual mechanism for computer-aided design
US6516451B1 (en) * 2000-07-05 2003-02-04 Clarence Wayne Patin Standards-integrated design software
US6622295B1 (en) * 2000-07-05 2003-09-16 Dupont Photomasks, Inc. Network-based photomask data entry interface and instruction generator for manufacturing photomasks
JP3696781B2 (ja) * 2000-09-05 2005-09-21 株式会社アマダ 板金部品の曲げ順及び曲げ金型提案装置及び提案方法
JP3610302B2 (ja) * 2000-12-28 2005-01-12 株式会社アマダ 板金加工作業のコスト見積もり作成方法
US6810295B1 (en) 2001-06-29 2004-10-26 Autodesk, Inc. Method and apparatus for simplified generation and displaying of cutout features of computer aided design (CAD) model
WO2003040841A1 (fr) * 2001-11-09 2003-05-15 Amada Company, Limited Systeme compose de traitement de toles
US6760640B2 (en) * 2002-03-14 2004-07-06 Photronics, Inc. Automated manufacturing system and method for processing photomasks
US7669167B2 (en) * 2002-07-30 2010-02-23 Photronics, Inc. Rule based system and method for automatically generating photomask orders by conditioning information from a customer's computer system
US6917856B2 (en) * 2002-04-12 2005-07-12 Mitsubishi Denki Kabushiki Kaisha Robot
FI112922B (fi) * 2002-06-14 2004-02-13 Finn Power Oy Menetelmä työstökonesolun ohjauksessa
JP3691032B2 (ja) * 2002-08-23 2005-08-31 ファナック株式会社 加工装置
EP1542918B1 (fr) * 2002-09-26 2016-11-16 Trumpf Maschinen Austria GmbH & CO. KG. Dispositif de préhension muni de moyens de détection d'une double alimentation et procédé d'utilisation correspondant
JP4630533B2 (ja) * 2002-10-23 2011-02-09 株式会社アマダ 曲げ加工機
US6819966B1 (en) 2003-12-06 2004-11-16 Paul E. Haeberli Fabrication of free form structures from planar materials
US8140306B2 (en) * 2004-04-27 2012-03-20 Autoform Engineering Gmbh Determination of process operations in order to describe forming processes on a forming part
US7398129B2 (en) * 2004-10-07 2008-07-08 Amada Company, Limited Representation of sheet metal part models
US7813901B2 (en) * 2004-10-25 2010-10-12 Amada Company, Limited Sketch generator for 3D sheet metal part models created by sheet metal part feature operations
ATE446534T1 (de) * 2004-10-25 2009-11-15 Alcoa Inc Virtuelle programmierung von formteilbahnen
US7158853B2 (en) * 2004-10-25 2007-01-02 Amada Company, Limited Pattern recognition for sheet metal part models
US7847807B2 (en) * 2004-12-22 2010-12-07 Hntb Holdings Ltd Geometry creation tool
US7711699B2 (en) * 2004-12-22 2010-05-04 Hntb Holdings Ltd. Method and system for presenting traffic-related information
US7849031B2 (en) * 2004-12-22 2010-12-07 Hntb Holdings Ltd. Optimizing traffic predictions and enhancing notifications
US7511634B2 (en) * 2004-12-22 2009-03-31 Htnb Corporation Retrieving and presenting dynamic traffic information
US7603191B2 (en) * 2005-02-16 2009-10-13 Idealab System and method for design of a component
JP2007075885A (ja) * 2005-09-16 2007-03-29 Fujitsu Ltd レーザ曲げ加工予測装置、レーザ曲げ加工予測プログラム、レーザ曲げ加工予測方法
JP5108260B2 (ja) * 2006-07-06 2012-12-26 株式会社アマダ 曲げ加工機金型レイアウトの活用方法およびその装置
US8078305B2 (en) 2007-03-23 2011-12-13 Siemens Product Lifecycle Management Software Inc. Numerical control arrangement
US20080281458A1 (en) * 2007-03-23 2008-11-13 Shengming Liu System and method for direct sheet metal unfolding
GB2444574B (en) * 2007-06-13 2008-11-12 Gregory Epps Bending sheet material
FR2935814B1 (fr) * 2008-09-05 2013-05-24 Chamacria Unipessoal Lda Procede de fabrication d'un objet en tole a l'aide d'une machine-outil a commande numerique
DE102009054112A1 (de) * 2009-11-20 2011-05-26 Kuka Roboter Gmbh Verfahren und Vorrichtung zur Planung und/oder Steuerung einer Roboterapplikation
US8886348B2 (en) 2011-05-25 2014-11-11 Kevin Bollendorf Fabrication and tracking
US20140019099A1 (en) * 2012-07-16 2014-01-16 Livermore Software Technology Corp Determination Of Failure In Sheet Metal Forming Simulation Using Isotropic Metal Failure Criteria
CN105451905B (zh) * 2013-08-09 2018-04-17 百超激光股份公司 折弯压力机
US10458890B2 (en) * 2013-12-20 2019-10-29 Thyssenkrupp Steel Europe Ag Method and device for determining the wear properties of coated flat products by means of bending
KR102149866B1 (ko) * 2015-09-29 2020-09-01 에이에스엠엘 네델란즈 비.브이. 리소그래피 시스템들과 같은 시스템들을 모델링하거나 시스템들의 예측 유지보수를 수행하는 방법들, 및 연계된 리소그래피 시스템들.
US10022872B2 (en) * 2015-10-06 2018-07-17 Mtm Robotics, Llc Self-contained modular manufacturing tool responsive to locally stored historical data
US10220516B2 (en) 2015-10-06 2019-03-05 Mtm Robotics, Llc System and method for self-contained independently controlled modular manufacturing tools
AT517966B1 (de) * 2016-02-22 2017-06-15 Ing Henning Scharsach Dipl Verfahren, Computerprogrammprodukt und Schwenkbiegemaschiene zum Biegen von Blechen
EP3311963A1 (fr) * 2016-10-21 2018-04-25 Bystronic Laser AG Cellule robotique positionnable, dispositif de fabrication comprenant un équipement de fabrication et une cellule robotique positionnable et procédé de fonctionnement d'une telle cellule robotique positionnable
US20180348742A1 (en) * 2017-06-01 2018-12-06 X Development Llc Planning and Adapting Projects Based on a Buildability Analysis
EP3648964A1 (fr) * 2017-07-06 2020-05-13 Bobst Mex Sa Machine de rainage, cylindre de rainage pour la machine de rainage et procédé de rainage de feuilles
EP3648908A1 (fr) * 2017-07-06 2020-05-13 Bobst Mex Sa Procédé de plissage de feuilles
DE102017213060A1 (de) * 2017-07-28 2019-01-31 Siemens Aktiengesellschaft Computerimplementiertes Verfahren und Vorrichtung zum automatischen Generieren von gekennzeichneten Bilddaten und Analysevorrichtung zum Überprüfen eines Bauteils
AT522518B1 (de) * 2019-05-10 2022-05-15 Trumpf Maschinen Austria Gmbh & Co Kg Verfahren und System zum Optimieren der Zusammensetzung von Biegewerkzeugen einer Biegemaschine
CN110298066B (zh) * 2019-05-15 2023-04-18 成都数模码科技有限公司 一种标准斜楔智能匹配方法
WO2021041419A1 (fr) * 2019-08-30 2021-03-04 X Development Llc Planification de robot à partir d'un graphe de définition de processus
CN112139320B (zh) * 2020-09-10 2023-03-03 中国航发贵州黎阳航空动力有限公司 一种螺旋状导管加工方法
US20220129710A1 (en) * 2020-10-26 2022-04-28 Hcl Technologies Limited System and method for classifying elements of a product
IT202100017384A1 (it) * 2021-07-01 2023-01-01 Blm Spa Procedimento di piegatura e macchina di piegatura per eseguire un procedimento di piegatura

Family Cites Families (63)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4369563A (en) * 1965-09-13 1983-01-25 Molins Limited Automated machine tool installation with storage means
US4309600A (en) * 1967-12-15 1982-01-05 Cincinnati Milacron Inc. Machine tool
US3654616A (en) * 1970-06-29 1972-04-04 Unimation Inc Program selection arrangement for manipulator apparatus
US3890552A (en) * 1972-12-29 1975-06-17 George C Devol Dual-armed multi-axes program controlled manipulators
GB1525044A (en) * 1975-02-12 1978-09-20 Bottomley A Tool for working sheet metal
US4260940A (en) * 1975-10-28 1981-04-07 Unimation, Inc. Programmable automatic assembly system
IT1122696B (it) * 1979-08-02 1986-04-23 Salvagnini Transferica Spa Perfezionamento nelle presse-piegatrici
JPS5841927B2 (ja) * 1979-10-25 1983-09-16 攻 牧野 プレス曲げ加工用工具
JPS57113116A (en) * 1980-12-30 1982-07-14 Fanuc Ltd Robot control system
JPH065486B2 (ja) * 1981-03-26 1994-01-19 株式会社安川電機 ロボットの軌跡制御方法
JPS5856003A (ja) * 1981-09-30 1983-04-02 Hitachi Ltd 工業用ロボツトの制御方法
US4501135A (en) * 1982-02-09 1985-02-26 L & F Industries Stress sensor for yield-point detection in a stretch press
CH655874A5 (de) * 1982-06-07 1986-05-30 Haemmerle Ag Maschf Werkzeugeinrichtung an einer abkantpresse.
JPS59114609A (ja) * 1982-12-22 1984-07-02 Hitachi Ltd ロボットの制御装置
JPS59189415A (ja) * 1983-04-13 1984-10-27 Hitachi Ltd 工業用ロボツトの動作教示方法および装置
CH654761A5 (fr) * 1984-02-03 1986-03-14 Beyeler Machines Sa Presse-plieuse dont le dispositif de pliage permet un controle continu de l'angle de pliage de la piece a plier.
US4658625A (en) * 1984-03-30 1987-04-21 Amada Company, Limited Bending machine and a die changing system for such bending machine
EP0162670B1 (fr) * 1984-05-19 1991-01-02 British Aerospace Public Limited Company Systèmes de travail et de fabrication industriels
US5146670A (en) * 1985-04-24 1992-09-15 The Boeing Company Profiling and deburring of workpieces
US4998206A (en) * 1988-07-29 1991-03-05 The Boeing Company Automated method and apparatus for fabricating sheet metal parts and the like using multiple manufacturing stations
US4942767A (en) * 1986-11-19 1990-07-24 Massachusetts Institute Of Technology Pressure transducer apparatus
JPS63140373A (ja) * 1986-12-02 1988-06-11 Oki Electric Ind Co Ltd Ｃａｄ／ｃａｍ装置における板金展開方法
US4745812A (en) * 1987-03-25 1988-05-24 The United States Of America As Represented By The Secretary Of The Army Triaxial tactile sensor
US5106290A (en) * 1987-04-14 1992-04-21 Northrop Corporation Assembly data model tool system
US4802357A (en) * 1987-05-28 1989-02-07 The Boeing Company Apparatus and method of compensating for springback in a workpiece
US4831549A (en) * 1987-07-28 1989-05-16 Brigham Young University Device and method for correction of robot inaccuracy
US5092645A (en) * 1987-09-18 1992-03-03 Wacoh Corporation Robotic gripper having strain sensors formed on a semiconductor substrate
DE3739173A1 (de) * 1987-11-19 1989-06-01 Feintool Int Holding Verfahren und vorrichtung zum biegen von werkstuecken
US4949277A (en) * 1988-03-09 1990-08-14 North American Philips Corporation Differential budding: method and apparatus for path planning with moving obstacles and goals
GB2211002B (en) * 1987-12-15 1992-01-08 Amada Co Ltd Device and method for controlling a manipulator for a plate bending machine
DE3902149C2 (de) * 1988-01-29 2000-05-18 Amada Co Biegeeinrichtung und Verfahren zum Positionieren von Werkstücken in einer Blechbiegeeinrichtung
IT1219110B (it) * 1988-03-15 1990-05-03 Prima Ind Spa Dispositivo manipolatore di pezzi di lamiera per una pressa piegatrice e procedimento per la fabbricazione di pezzi profilati di lamiera mediante piegatura alla pressa
US5047916A (en) * 1988-03-25 1991-09-10 Kabushiki Kaisha Toshiba Method and apparatus of free space enumeration for collision avoidance
FR2630358B1 (fr) * 1988-04-21 1993-12-10 Picot Sa Procede et dispositif pour le controle du retour elastique, lors du cintrage d'un element allonge tel que tube
AT389829B (de) * 1988-05-03 1990-02-12 Haemmerle Ag Verfahren zum biegen von blechstuecken mit hilfe einer biegeeinrichtung
IT1219302B (it) * 1988-05-16 1990-05-03 Prima Ind Spa Macchina per la fabbricazione di pezzi di lamiera piegata
US5031441A (en) * 1988-07-29 1991-07-16 The Boeing Company Two arm robot
SE506952C2 (sv) * 1988-08-05 1998-03-09 Amada Co Ltd Sätt och anordning för inställning av en bockningsprocess, och ett sätt att iordningställa bockningsdata
DE3830488A1 (de) * 1988-09-08 1990-03-15 Fastenrath Fasti Werk Elektronisches werkzeugerkennungssystem fuer gesenkbiegepressen
US4947666A (en) * 1988-09-16 1990-08-14 The Boeing Company Method and apparatus for bending an elongate workpiece
AT392023B (de) * 1988-11-21 1991-01-10 Haemmerle Ag Verfahren zum biegen von blechstuecken
IT1224044B (it) * 1988-12-29 1990-09-26 Prima Ind Spa Pressa piegatrice di precisione per pezzi di lamiera lunghi
US5081763A (en) * 1989-05-30 1992-01-21 The Boeing Company Automated drilling station
JPH0832341B2 (ja) * 1989-08-31 1996-03-29 株式会社小松製作所 プレスブレーキの制御装置
US5307282A (en) * 1989-09-22 1994-04-26 Hewlett-Packard Company Method of computer-aided prediction of collisions between objects including fabrication tools and parts to be fabricated
JP2773917B2 (ja) * 1989-09-22 1998-07-09 株式会社アマダ ベンディング装置のワーク位置決め装置
JP2727241B2 (ja) * 1989-09-26 1998-03-11 株式会社小松製作所 プレスブレーキシステムのワーク位置決め装置および方法
US5089970A (en) * 1989-10-05 1992-02-18 Combustion Engineering, Inc. Integrated manufacturing system
JP2723994B2 (ja) * 1989-10-06 1998-03-09 株式会社アマダ ベンディング装置のゲージング処理方法
US5009098A (en) * 1989-11-27 1991-04-23 Machinefabriek Wila B.V. Press and curve-forming means therefor
WO1991008869A1 (fr) * 1989-12-18 1991-06-27 Amada Company, Limited Dispositif de commande d'un manipulateur robotique pour une installation de pliage de plaques de metal
US5150288A (en) * 1989-12-18 1992-09-22 Honda Giken Kogyo Kabushiki Kaisha Production management system and method of transmitting data
IT1237750B (it) * 1989-12-29 1993-06-15 Prima Ind Spa Procedimento di piegatura di una lamiera
US5329597A (en) * 1990-02-23 1994-07-12 Amada Company, Ltd. Device and method for measuring angles of a work
US5036694A (en) * 1990-07-02 1991-08-06 The Boeing Company Method and apparatus for bending the flange of a workpiece
US5088181A (en) * 1990-10-09 1992-02-18 The Boeing Company Sheet metal part machining system
US5325468A (en) * 1990-10-31 1994-06-28 Sanyo Electric Co., Ltd. Operation planning system for robot
JP2661395B2 (ja) * 1991-04-09 1997-10-08 ダイキン工業株式会社 プレスブレーキシステムにおけるワーク位置決め方法およびその装置
US5377516A (en) * 1992-03-19 1995-01-03 Laser Products, Inc. Method and apparatus for bending sheet stock
JP3181382B2 (ja) * 1992-07-15 2001-07-03 株式会社アマダ 折曲げ加工機のワーク位置決め装置
US5390283A (en) * 1992-10-23 1995-02-14 North American Philips Corporation Method for optimizing the configuration of a pick and place machine
US5315522A (en) * 1992-11-06 1994-05-24 Kenneth A. Kauffman Table bender controller
IT1260677B (it) * 1993-07-29 1996-04-22 Antonio Codatto Manipolatore per la movimentazione di lastre, particolarmente pannellidi lamiera, nei confronti di una macchina operatrice, quale una pressapiegatrice.

Also Published As

Publication number Publication date
DE69535946D1 (de) 2009-06-10
EP1684140B1 (fr) 2009-04-29
DE69535954D1 (de) 2009-06-25
EP1657607B1 (fr) 2009-05-13
EP1681607A2 (fr) 2006-07-19
DE69535687T2 (de) 2009-01-02
EP1684140A3 (fr) 2007-03-07
DE69535687D1 (de) 2008-02-21
EP1657607A3 (fr) 2007-03-07
US5969973A (en) 1999-10-19
EP1253496A1 (fr) 2002-10-30
EP1681607A3 (fr) 2006-08-02
EP1657607A2 (fr) 2006-05-17
EP1684140A2 (fr) 2006-07-26
EP1681607B1 (fr) 2008-01-09

Similar Documents

Publication Publication Date Title
EP1253496B1 (fr) Dispositif de pliage comprenant un système de contrôle servant à générer et à exécuter un plan de pliage de tôles métalliques
EP0744046B1 (fr) Systeme intelligent servant a generer et a executer un plan de pliage de toles metalliques
JP4413423B2 (ja) 板金曲げ作業に関する段取り計画を作成する方法
Gupta et al. Automated process planning for sheet metal bending operations
EP0925157B1 (fr) Dispositif et procede pour integrer un systeme de fabrication intelligent a un systeme expert de planification et de pliage de tole
JPH09509513A (ja) ロボットの運動を計画し制御する方法
Meddahi et al. API based graphical simulation of robotized sites
Wang et al. Using features and their constraints to aid process planning of sheet metal parts
De Vin et al. Tolerancing and sheet bending in small batch part manufacturing
Pashkevich et al. Multiobjective optimisation of robot location in a workcell using genetic algorithms
Roßgoderer et al. A concept for automatical layout generation
JP3549842B2 (ja) 高知能製造システムをエキスパート曲げ計画作成システムと統合する方法及び装置
US20230109876A1 (en) Information processing apparatus, information processing method, and method of manufacturing products
WO2022230544A1 (fr) Dispositif, procédé et programme de conception d'équipement robotisé
Dolgui et al. Enhanced Optimisation Techniques for Off-line Programming of Laser Cutting Robots
Wohlke A simulation-based grasp planning system for multifinger robot hands

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

AC Divisional application: reference to earlier application

Ref document number: 744046

Country of ref document: EP

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): DE

17P Request for examination filed

Effective date: 20030128

AKX Designation fees paid

Designated state(s): DE

17Q First examination report despatched

Effective date: 20030730

GRAP Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOSNIGR1

RTI1 Title (correction)

Free format text: BENDING APPARATUS COMPRISING A CONTROL MEANS BEING CONFIGURED TO GENERATE AND EXECUTE A SHEETMETAL BENDING PLAN

GRAS Grant fee paid

Free format text: ORIGINAL CODE: EPIDOSNIGR3

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AC Divisional application: reference to earlier application

Ref document number: 0744046

Country of ref document: EP

Kind code of ref document: P

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): DE

REF Corresponds to:

Ref document number: 69534977

Country of ref document: DE

Date of ref document: 20060608

Kind code of ref document: P

PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

26N No opposition filed

Effective date: 20070206

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: DE

Payment date: 20131127

Year of fee payment: 19

REG Reference to a national code

Ref country code: DE

Ref legal event code: R119

Ref document number: 69534977

Country of ref document: DE

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: DE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20150602